agora inbox for [email protected]  

help / color / mirror / Atom feed

HTTP Frontend? (and a brief thought on materialized views)
8673+ messages / 6 participants
[nested] [flat]

* HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-29 14:37  Dobes Vandermeer <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-29 14:37 UTC (permalink / raw)
  To: pgsql-hackers

Hi guys,

Something from Josh's recent blog post about summer of code clicked with me
- the HTTP / SQL concept.

It was something I'd been thinking about earlier, how people really like
HTTP APIs and this is one of the drivers behind adoption of  some "NoSQL"
databases out there.

Some things that I think are great about NoSQL that are missing in
PostgreSQL are:

   1. The "reduce" function which could be modeled as a kind of
   materialized view with an index on it.  With materialized views and the
   ability to pull fields from JSON and XML data structures you could easily
   do a "NoSQL" database inside of PostgreSQL by saving the document as a big
   string and then using materialized views to have all kinds of derived data
   from those documents.  Sounds cool on paper, anyway.
   2. HTTP RESTful API.  This is obviously not as useful as a good admin
   tool like pgAdmin or a fast binary protocol like we have now but it's way
   more buzzword compliant and would come in handy sometimes.  With CouchDB I
   was able to allow "public" data in the database to be accessed directly
   from the browser using JSONP and/or a simple HTTP proxy in the server
   instead of doing any of that work within the app server.  So, that saves a
   step somewhere.  With some basic support for stored procedures and serving
   files directly via this HTTP thing you could potentially eliminate the app
   server for public, read-only parts of a site/application.
   3. The perception of extremely low latency and/or high throughput - like
   fetching a row in less than 1ms or whatever.  I don't know the exact
   numbers I just know they have to be insanely low (latency) or impressively
   high (throughput).  We could use PostgreSQL as a query cache to accelerate
   your MySQL!  Something like that :-P.
   4. Rebelliousness against "the man".  In our case SQL can't be "the
   man", but we can find something PostgreSQL isn't and position against that.

Anyway, 1&2 seem inspiring to me and I'd love to help out with both.  3&4
seem more like marketing tasks of some sort...

I think I could probably start hacking up an HTTP API of some sort, I
wasn't able to log into the PostgreSQL Wiki so I created a page with some
notes here:

http://dobesv.com/docs/postgresql-http-api.html

For materialized views I think that can be a bit of a can of worms,
especially if you want to do incremental updates of some sort because you
have to figure out what rows need to be recalculated when you update a row
in a "source" table, and how best to update them.  But if someone thinks
they know how I can assist in implementation.

Anyway, I hope I can help AND that I posted this in the correct mailing
list!

Cheers,

Dobes


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-29 15:04  Andrew Dunstan <[email protected]>
  parent: Dobes Vandermeer <[email protected]>
  0 siblings, 3 replies; 8673+ messages in thread

From: Andrew Dunstan @ 2012-03-29 15:04 UTC (permalink / raw)
  To: Dobes Vandermeer <[email protected]>; +Cc: pgsql-hackers



On 03/29/2012 10:37 AM, Dobes Vandermeer wrote:
> Hi guys,
>
> Something from Josh's recent blog post about summer of code clicked 
> with me - the HTTP / SQL concept.
>
> It was something I'd been thinking about earlier, how people really 
> like HTTP APIs and this is one of the drivers behind adoption of  some 
> "NoSQL" databases out there.
>
> Some things that I think are great about NoSQL that are missing in 
> PostgreSQL are:
>
>  1. The "reduce" function which could be modeled as a kind of
>     materialized view with an index on it.  With materialized views
>     and the ability to pull fields from JSON and XML data structures
>     you could easily do a "NoSQL" database inside of PostgreSQL by
>     saving the document as a big string and then using materialized
>     views to have all kinds of derived data from those documents.
>      Sounds cool on paper, anyway.
>  2. HTTP RESTful API.  This is obviously not as useful as a good admin
>     tool like pgAdmin or a fast binary protocol like we have now but
>     it's way more buzzword compliant and would come in handy
>     sometimes.  With CouchDB I was able to allow "public" data in the
>     database to be accessed directly from the browser using JSONP
>     and/or a simple HTTP proxy in the server instead of doing any of
>     that work within the app server.  So, that saves a step somewhere.
>      With some basic support for stored procedures and serving files
>     directly via this HTTP thing you could potentially eliminate the
>     app server for public, read-only parts of a site/application.
>  3. The perception of extremely low latency and/or high throughput -
>     like fetching a row in less than 1ms or whatever.  I don't know
>     the exact numbers I just know they have to be insanely low
>     (latency) or impressively high (throughput).  We could use
>     PostgreSQL as a query cache to accelerate your MySQL!  Something
>     like that :-P.
>  4. Rebelliousness against "the man".  In our case SQL can't be "the
>     man", but we can find something PostgreSQL isn't and position
>     against that.
>
> Anyway, 1&2 seem inspiring to me and I'd love to help out with both. 
>  3&4 seem more like marketing tasks of some sort...
>
> I think I could probably start hacking up an HTTP API of some sort, I 
> wasn't able to log into the PostgreSQL Wiki so I created a page with 
> some notes here:
>
> http://dobesv.com/docs/postgresql-http-api.html
>
> For materialized views I think that can be a bit of a can of worms, 
> especially if you want to do incremental updates of some sort because 
> you have to figure out what rows need to be recalculated when you 
> update a row in a "source" table, and how best to update them.  But if 
> someone thinks they know how I can assist in implementation.
>
> Anyway, I hope I can help AND that I posted this in the correct 
> mailing list!
>


1. I've been in discussion with some people about adding simple JSON 
extract functions. We already have some (i.e. xpath()) for XML.

2. You might find htsql <http://htsql.org/; interesting.


cheers

andrew



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-29 19:57  Daniel Farina <[email protected]>
  parent: Andrew Dunstan <[email protected]>
  2 siblings, 2 replies; 8673+ messages in thread

From: Daniel Farina @ 2012-03-29 19:57 UTC (permalink / raw)
  To: Andrew Dunstan <[email protected]>; +Cc: Dobes Vandermeer <[email protected]>; pgsql-hackers

On Thu, Mar 29, 2012 at 8:04 AM, Andrew Dunstan <[email protected]> wrote:
 1. I've been in discussion with some people about adding simple JSON extract
> functions. We already have some (i.e. xpath()) for XML.
>
> 2. You might find htsql <http://htsql.org/; interesting.

My colleagues and myself have thought about this quite a lot.  Even
without changing any of FEBE's semantics (so  far, I think they're
basically fine, with the glaring problem of hidden session state) I
think it may be a good time to experiment in supporting SPDY (clearly,
9.3+), which solves some of the problems of HTTP that make it pretty
unusable for interactive database workload.

For a long time, it looked like SPDY would become the IETF's HTTP 2.0.
 But the future is still clouded, as just recently Microsoft
introduced their own mostly-compatible design.  So that's a bit scary.

More concretely, here's what I really want.  The first two concerns
are almost entirely technical, the latter two social, in that they
rely on notion of "common" or "standard':

* A standard frame extension format.  For example, last I checked
Postgres-XC, it required snapshot information to be passed, and it'd
be nice that instead of having to hack the protocol that they could
attach an X-Snapshot-Info header, or whatever. This could also be a
nice way to pass out-of-band hint information of all sorts.

* Something similar to the HTTP "Host" header, so that one can route
to databases without having to conflate database identity with the
actual port being connected to.  Yes, theoretically it can be done
with weird startup packet gyrations, but that is firmly in the "weird"
category.

* HTTP -- and *probably* its hypothetical progeny -- are more common
than FEBE packets, and a lot of incidental complexity of writing
routers is reduced by the commonality of routing HTTP traffic.

* Protocol compression -- but a bit of sand in the gears is *which*
compression -- for database workloads, the performance of zlib can be
a meaningful bottleneck.

Lastly, a case that can not as easily be fixed without some more
thinking is leveraging caching semantics of HTTP.   think people would
really, really like that, if they could get away from having to
hand-roll their own cache regeneration in common cases.

-- 
fdr



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-29 19:59  Daniel Farina <[email protected]>
  parent: Daniel Farina <[email protected]>
  1 sibling, 1 reply; 8673+ messages in thread

From: Daniel Farina @ 2012-03-29 19:59 UTC (permalink / raw)
  To: Andrew Dunstan <[email protected]>; +Cc: Dobes Vandermeer <[email protected]>; pgsql-hackers

On Thu, Mar 29, 2012 at 12:57 PM, Daniel Farina <[email protected]> wrote:
D'oh, I munged the order.

More technical concerns:
> * Protocol compression -- but a bit of sand in the gears is *which*
> compression -- for database workloads, the performance of zlib can be
> a meaningful bottleneck.

> * Something similar to the HTTP "Host" header, so that one can route
> to databases without having to conflate database identity with the
> actual port being connected to.  Yes, theoretically it can be done
> with weird startup packet gyrations, but that is firmly in the "weird"
> category.

Socialish (but no less important):

> * A standard frame extension format.  For example, last I checked
> Postgres-XC, it required snapshot information to be passed, and it'd
> be nice that instead of having to hack the protocol that they could
> attach an X-Snapshot-Info header, or whatever. This could also be a
> nice way to pass out-of-band hint information of all sorts.
>
>
> * HTTP -- and *probably* its hypothetical progeny -- are more common
> than FEBE packets, and a lot of incidental complexity of writing
> routers is reduced by the commonality of routing HTTP traffic.


-- 
fdr



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 01:25  Dobes Vandermeer <[email protected]>
  parent: Andrew Dunstan <[email protected]>
  2 siblings, 1 reply; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-30 01:25 UTC (permalink / raw)
  To: pgsql-hackers

On Thu, Mar 29, 2012 at 11:04 PM, Andrew Dunstan <[email protected]>wrote:

>
> On 03/29/2012 10:37 AM, Dobes Vandermeer wrote:
>
>> Hi guys,
>>
>> Something from Josh's recent blog post about summer of code clicked with
>> me - the HTTP / SQL concept.
>>
>>

>  1. I've been in discussion with some people about adding simple JSON
> extract functions. We already have some (i.e. xpath()) for XML.
>
> 2. You might find htsql <http://htsql.org/; interesting.
>

As a reference, or should we just bundle / integrate that with PostgreSQL
somehow?


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 01:36  Daniel Farina <[email protected]>
  parent: Dobes Vandermeer <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Daniel Farina @ 2012-03-30 01:36 UTC (permalink / raw)
  To: Dobes Vandermeer <[email protected]>; +Cc: pgsql-hackers

On Thu, Mar 29, 2012 at 6:25 PM, Dobes Vandermeer <[email protected]> wrote:
>> 2. You might find htsql <http://htsql.org/; interesting.
>
>
> As a reference, or should we just bundle / integrate that with PostgreSQL
> somehow?

It's a totally different language layer without wide-spread popularity
and, as of last year, still figuring out how to model updates and
transactions, with a Python dependency.  It is good work, but to build
it in and to commit maintaining it seems very premature, to my eyes.
Also, it would tie someone -- possibly the htsql maintainers -- to a
upgrade and maintenance policy not unlike Postgres itself, and that is
a very expensive proposition while they are still figuring things out.

-- 
fdr



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 01:37  Dobes Vandermeer <[email protected]>
  parent: Daniel Farina <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-30 01:37 UTC (permalink / raw)
  To: Daniel Farina <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 3:59 AM, Daniel Farina <[email protected]> wrote:

> On Thu, Mar 29, 2012 at 12:57 PM, Daniel Farina <[email protected]>
> wrote:

More technical concerns:
> > * Protocol compression -- but a bit of sand in the gears is *which*
> > compression -- for database workloads, the performance of zlib can be
> > a meaningful bottleneck.
>

I think if performance is the issue, people should use the native protocol.
 This HTTP thing should be more of a RAD / prototyping thing, I think.  So
people can be in their comfort zone when talking to the server.


> * Something similar to the HTTP "Host" header, so that one can route
> > to databases without having to conflate database identity with the
> > actual port being connected to.  Yes, theoretically it can be done
> > with weird startup packet gyrations, but that is firmly in the "weird"
> > category.
>

Isn't the URL good enough (/databases/<dbname>) or are you talking about
having some some of "virtual host" setup where you have multiple sets of
databases available on the same port?


Socialish (but no less important):
>
> > * A standard frame extension format.  For example, last I checked
> > Postgres-XC, it required snapshot information to be passed, and it'd
> > be nice that instead of having to hack the protocol that they could
> > attach an X-Snapshot-Info header, or whatever. This could also be a
> > nice way to pass out-of-band hint information of all sorts.
>

I am sorry to admit I don't understand the terms "frame extension format"
or "Postgres-XC" in this paragraph ... help?


> > * HTTP -- and *probably* its hypothetical progeny -- are more common
>  > than FEBE packets, and a lot of incidental complexity of writing
> > routers is reduced by the commonality of routing HTTP traffic.
>

This is an interesting comment.  I'm not sure how to test whether an HTTP
based protocol will be better supported than a proprietary one, but I think
we have enough other reasons that we can move forward.  Well we have the
reason that there's some kind of love affair with HTTP based protocols
going on out there ... might as well ride the wave while it's still rising
(I hope).

As for SPDY I can see how it may be helpful but as it is basically a
different way to send HTTP requests (from what I understand) the migration
to SPDY is mainly a matter of adding support for it to whatever HTTP
library is used.

Anyone have a thought on whether, for the HTTP server itself, it should be
integrated right into the PostgreSQL server itself?  Or would it be better
to have a separate process that proxies requests to PostgreSQL using the
existing protocol?  Is there an API that can be used in both cases
semi-transparently (i.e. the functions have the same name when linked right
in, or when calling via a socket)?

Cheers,

Dobes


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 02:12  Dobes Vandermeer <[email protected]>
  parent: Daniel Farina <[email protected]>
  1 sibling, 0 replies; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-30 02:12 UTC (permalink / raw)
  To: Daniel Farina <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 3:57 AM, Daniel Farina <[email protected]> wrote:

> On Thu, Mar 29, 2012 at 8:04 AM, Andrew Dunstan <[email protected]>
> wrote:
>


> Lastly, a case that can not as easily be fixed without some more
> thinking is leveraging caching semantics of HTTP.   think people would
> really, really like that, if they could get away from having to
> hand-roll their own cache regeneration in common cases.
>

I think this could be an interesting possibility.  I wonder if the cost of
a round-trip makes the cost of sending the actual data (vs a 304 response)
irrelevant - as long as PostgreSQL is caching effectively internally it's
possible it can send back the actual content as fast as it can calculate
the ETag for it, so doing an extra query to check for changes could
possibly slow things down, or at least eliminate the benefit.  Probably
worth trying, though.

Supporting this "automagically" would require some kind of generic
algorithm for calculating the Last-Modifed time or ETag for a given query.
 As a default we may be able to just fall back on some internal global
value that is guaranteed to change if the database has changed (I think the
WAL files have some kind of serial number system we might use) so at the
very least you could send back a 304 Not Modified if literally nothing in
the database has changed.  Narrowing that down to specific table timestamps
might be possible, too, for simple queries.  It depends what data is
already available, I wouldn't want to add any extra book keeping for it.

A more pragmatic may be to have the HTTP request include SQL code to
generate an ETag or Last-Modified value to test with; the app could run
that first and it would be used for caching.  Something like
calcLastModified=max(modified_date) on the query string.


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 02:55  Daniel Farina <[email protected]>
  parent: Dobes Vandermeer <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Daniel Farina @ 2012-03-30 02:55 UTC (permalink / raw)
  To: Dobes Vandermeer <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; pgsql-hackers

On Thu, Mar 29, 2012 at 6:37 PM, Dobes Vandermeer <[email protected]> wrote:
> On Fri, Mar 30, 2012 at 3:59 AM, Daniel Farina <[email protected]> wrote:
>>
>> On Thu, Mar 29, 2012 at 12:57 PM, Daniel Farina <[email protected]>
>> wrote:
>>
>> More technical concerns:
>> > * Protocol compression -- but a bit of sand in the gears is *which*
>> > compression -- for database workloads, the performance of zlib can be
>> > a meaningful bottleneck.
>
>
> I think if performance is the issue, people should use the native protocol.
>  This HTTP thing should be more of a RAD / prototyping thing, I think.  So
> people can be in their comfort zone when talking to the server.

No. I do not think so. I think a reasonable solution (part of what MS
is actually proposing to the IETF) is to make compression optional, or
have clients support an LZ77 family format like Snappy.  I get the
impression that SPDY is waffling a little on this fact, it mandates
compression, and definitely zlib, but is less heavy handed about
pushing, say Snappy.  Although I can understand why a
Google-originated technology probably doesn't want to push another
Google-originated implementation so hard, it would have been
convenient for me for Snappy to have become a more common format.

> Isn't the URL good enough (/databases/<dbname>) or are you talking about
> having some some of "virtual host" setup where you have multiple sets of
> databases available on the same port?

Virtual hosts. Same port.

>> > * A standard frame extension format.  For example, last I checked
>> > Postgres-XC, it required snapshot information to be passed, and it'd
>> > be nice that instead of having to hack the protocol that they could
>> > attach an X-Snapshot-Info header, or whatever. This could also be a
>> > nice way to pass out-of-band hint information of all sorts.
>
>
> I am sorry to admit I don't understand the terms "frame extension format" or
> "Postgres-XC" in this paragraph ... help?

I'm being vague.  Postgres-XC is a project to provide a shared-nothing
sync-rep cluster for Postgres.  My last understanding of it is that it
needed to pass snapshot information between nodes, and FEBE was
expanded to make room for this, breaking compatibility, as well as
probably being at least a small chore.  It'd be nice if it wasn't
necessary to do that and they had a much easier path to multiplex
additional information into the connection.

For my own purposes, I'm intensely interest in lacing the connection with:

* EXPLAIN ANALYZE returns when the query has already run, getting both
the actual timings *and* the results to the client.

* Partition IDs, whereby you can find the right database and
(potentially!) even influence how the queries are scoped to a tenant

* Read-only vs. Write workload: As is well established, it's hard to
know a-priori if a query is going to do a write.  Fine. Let the client
tag it, signal an error if something is wrong.

Yes, theoretically all these features -- or just a general
multiplexing scheme -- can be added to FEBE, but if we're even going
to consider such an upheaval, maybe we can get *lot* more bang for our
buck by trying to avoid being unnecessarily different from the most
common application-level protocol in existence, causing extraneous
work for router and proxy authors.  Notably, a vanilla Postgres
database knows nothing about these extension headers.

>> > * HTTP -- and *probably* its hypothetical progeny -- are more common
>> > than FEBE packets, and a lot of incidental complexity of writing
>> > routers is reduced by the commonality of routing HTTP traffic.
>
> This is an interesting comment.  I'm not sure how to test whether an HTTP
> based protocol will be better supported than a proprietary one, but I think
> we have enough other reasons that we can move forward.  Well we have the
> reason that there's some kind of love affair with HTTP based protocols going
> on out there ... might as well ride the wave while it's still rising (I
> hope).

At its core, what may be growing unnecessary is FEBE's own mechanism
for delimiting messages. All the other protocol actions -- such as
shipping Binds, Executes, Describes, et al, are not going to be
obsoleted or even changed by laying web-originated technologies under
FEBE.

Consider the wealth of projects, products, and services that filter
HTTP vs FEBE, many quite old, now.  In my mind, "wave" might be better
rendered "tsunami".  The very real problem, as I see it, is that
classic, stateless HTTP would be just too slow to be practical.

> As for SPDY I can see how it may be helpful but as it is basically a
> different way to send HTTP requests (from what I understand) the migration
> to SPDY is mainly a matter of adding support for it to whatever HTTP library
> is used.

I think SPDY or like-protocols (there's only one other I can think of,
the very recently introduced and hilariously branded "S&M" from
Microsoft.  It's memorable, at least) are the only things that appear
to give a crisp treatment to interactive, stateful workloads involving
back-and-forth between client and server with multiplexing, fixing
some problems with the hacks in HTTP-land from before.

> Anyone have a thought on whether, for the HTTP server itself, it should be
> integrated right into the PostgreSQL server itself?  Or would it be better
> to have a separate process that proxies requests to PostgreSQL using the
> existing protocol?  Is there an API that can be used in both cases
> semi-transparently (i.e. the functions have the same name when linked right
> in, or when calling via a socket)?

If SPDY/HTTP2.0 were more common/existent in the latter case, I'd
advocate for swallowing it and making it just part of the monolithic
system.  But it is still a time of transition, and jumping the gun on
it would be expensive.

-- 
fdr



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 05:55  Dobes Vandermeer <[email protected]>
  parent: Daniel Farina <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-30 05:55 UTC (permalink / raw)
  To: Daniel Farina <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 10:55 AM, Daniel Farina <[email protected]> wrote:

> On Thu, Mar 29, 2012 at 6:37 PM, Dobes Vandermeer <[email protected]>
> wrote:
> > On Fri, Mar 30, 2012 at 3:59 AM, Daniel Farina <[email protected]>
> wrote:
> >>
> >> On Thu, Mar 29, 2012 at 12:57 PM, Daniel Farina <[email protected]>
> >> wrote:
> >>
> >> More technical concerns:
> >> > * Protocol compression -- but a bit of sand in the gears is *which*
> >> > compression -- for database workloads, the performance of zlib can be
> >> > a meaningful bottleneck.
> >
> > I think if performance is the issue, people should use the native
> protocol.
>
> No. I do not think so. I think a reasonable solution (part of what MS
> is actually proposing to the IETF) is to make compression optional, or
> have clients support an LZ77 family format like Snappy.  I get the
> impression that SPDY is waffling a little on this fact, it mandates
> compression, and definitely zlib, but is less heavy handed about
> pushing, say Snappy.  Although I can understand why a
> Google-originated technology probably doesn't want to push another
> Google-originated implementation so hard, it would have been
> convenient for me for Snappy to have become a more common format.
>
> > Isn't the URL good enough (/databases/<dbname>) or are you talking about
> > having some some of "virtual host" setup where you have multiple sets of
> > databases available on the same port?
>
> Virtual hosts. Same port.
>

In that case, the frontend would not be tied to a specific PostgreSQL
server, then?  I think initially this might complicate things a bit, and
you could solve it by putting an HTTP proxy in front to do the virtual
hosts for you.


> >> > * A standard frame extension format.  For example, last I checked
> >> > Postgres-XC, it required snapshot information to be passed, and it'd
> >> > be nice that instead of having to hack the protocol that they could
> >> > attach an X-Snapshot-Info header, or whatever. This could also be a
> >> > nice way to pass out-of-band hint information of all sorts.
> >
> >
> > I am sorry to admit I don't understand the terms "frame extension
> format" or
> > "Postgres-XC" in this paragraph ... help?
>
> It'd be nice if it wasn't
> necessary to do that and they had a much easier path to multiplex
> additional information into the connection.
>

Ah, I get it - you want a way to add some extra information to the protocol
in a backwards compatible way.  HTTP (and SPDY) provides a "standard" way
to do that.  Makes sense.


>
> For my own purposes, I'm intensely interest in lacing the connection with:
>
> * EXPLAIN ANALYZE
> * Partition IDs
> * Read-only vs. Write workload


I'll make a note of these and hash out the details a bit more once there's
something working to add them to.


> > As for SPDY I can see how it may be helpful but as it is basically a

> different way to send HTTP requests
>
> I think SPDY or like-protocols [...] give a crisp treatment to
> interactive, stateful workloads involving
> back-and-forth between client and server with multiplexing, fixing
> some problems with the hacks in HTTP-land from before.
>

It sounds like at some level you're really talking about replacing the
built-in protocol with SPDY because SPDY is possibly a better baseline than
updating the existing protocol.  That's an interesting idea, I think this
project could evolve in that direction if there's demand for it.


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 08:30  Daniel Farina <[email protected]>
  parent: Dobes Vandermeer <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Daniel Farina @ 2012-03-30 08:30 UTC (permalink / raw)
  To: Dobes Vandermeer <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; pgsql-hackers

On Thu, Mar 29, 2012 at 10:55 PM, Dobes Vandermeer <[email protected]> wrote:
>> Virtual hosts. Same port.
>
> In that case, the frontend would not be tied to a specific PostgreSQL
> server, then?  I think initially this might complicate things a bit, and you
> could solve it by putting an HTTP proxy in front to do the virtual hosts for
> you.

I think these problems are treatable, as you mention, and if there is
to be any value at all by using mostly off-the-shelf components
(which, at this time, are rarer for the new-generation stuff).  That's
the draw, for me.

> Ah, I get it - you want a way to add some extra information to the protocol
> in a backwards compatible way.  HTTP (and SPDY) provides a "standard" way to
> do that.  Makes sense.
>
> I'll make a note of these and hash out the details a bit more once there's
> something working to add them to.

A lot of them are old ideas, but it would be nice to encourage
experimentation by getting over some of the
small-matter-of-programming and backwards-compatibility issues.

>> I think SPDY or like-protocols [...] give a crisp treatment to
>> interactive, stateful workloads involving
>>
>> back-and-forth between client and server with multiplexing, fixing
>> some problems with the hacks in HTTP-land from before.
>
> It sounds like at some level you're really talking about replacing the
> built-in protocol with SPDY because SPDY is possibly a better baseline than
> updating the existing protocol.  That's an interesting idea, I think this
> project could evolve in that direction if there's demand for it.

If only so there is a smaller set of arbitrary decisions to make about
how to delimit messages...but if SPDY doesn't get widely deployed, or
exacts an unacceptable performance penalty, it is game over.  The
worst possible outcome is the thing that becomes common also is
impractical for session-oriented sequential SQL execution, but I am
hopeful that given the use cases driving this standardization process
that this is less likely to happen.  FEBE's framing/message format
doesn't seem like an area where people are itching to try wild and
crazy changes unique to Postgres (maybe someone has...) so using a
more commonly seen delimitation format seems like a concession that
opens a lot of more useful interesting doors.

This discussion is somewhat premature because HTTP 2.0 isn't
standardized, nor has SPDY become a wide-spread defacto format
(although the percentage of well-known web-infrastructure projects
implementing it has grown both impressively both recently and
overall), and there are not even straw-man bandwidth/latency
measurements (in terms of vs. FEBE), but as long as someone is
bringing up HTTP, I thought it worth discussing in a little more
depth, because it's something I poll regularly mentally, looking for a
sign that It's Time.  It wasn't ready enough for me to start a thread,
but clearly I couldn't quite resist replying to one...

-- 
fdr



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 13:57  Merlin Moncure <[email protected]>
  parent: Andrew Dunstan <[email protected]>
  2 siblings, 2 replies; 8673+ messages in thread

From: Merlin Moncure @ 2012-03-30 13:57 UTC (permalink / raw)
  To: Andrew Dunstan <[email protected]>; +Cc: Dobes Vandermeer <[email protected]>; pgsql-hackers

On Thu, Mar 29, 2012 at 10:04 AM, Andrew Dunstan <[email protected]> wrote:
> 1. I've been in discussion with some people about adding simple JSON extract
> functions. We already have some (i.e. xpath()) for XML.

I've built a couple of applications that push data in and out of xml
via manual composition going out and xpath coming in.  TBH, I found
this to be a pretty tedious way of developing a general application
structure and a couple of notches down from the more sql driven
approach.  Not that jsonpath/xpath aren't wonderful functions -- but I
thing for general information passing there's a better way.

Your json work is a great start in marrying document level database
features with a relational backend.  My take is that storing rich data
inside the database in json format, while tempting, is generally a
mistake.  Unless the document is black box it should be decomposed and
stored relationally and marked back up into a document as it goes out
the door.  This is why brevity and flexibility of syntax is so
important when marshaling data in and out of transport formats.  It
encourages people to take the right path and get the best of both
worlds -- a rich backend with strong constraints that can natively
speak such that writing data driven web services is easy.

What I'm saying is that jsonpath probably isn't the whole story:
another way of bulk moving json into native backend structures without
parsing would also be very helpful.  For example, being able to cast a
json document into a record or a record array would be just amazing.

merlin



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 14:31  Andrew Dunstan <[email protected]>
  parent: Merlin Moncure <[email protected]>
  1 sibling, 0 replies; 8673+ messages in thread

From: Andrew Dunstan @ 2012-03-30 14:31 UTC (permalink / raw)
  To: Merlin Moncure <[email protected]>; +Cc: Dobes Vandermeer <[email protected]>; pgsql-hackers



On 03/30/2012 09:57 AM, Merlin Moncure wrote:
> What I'm saying is that jsonpath probably isn't the whole story: 
> another way of bulk moving json into native backend structures without 
> parsing would also be very helpful. For example, being able to cast a 
> json document into a record or a record array would be just amazing.

It should be possible. After all, hstore's populate_record() does that. 
If we married that logic up with the json parser we should be half way 
there.

cheers

andrew



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 14:55  Dobes Vandermeer <[email protected]>
  parent: Daniel Farina <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-30 14:55 UTC (permalink / raw)
  To: Daniel Farina <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 4:30 PM, Daniel Farina <[email protected]> wrote:

> On Thu, Mar 29, 2012 at 10:55 PM, Dobes Vandermeer <[email protected]>
> wrote:
> >> Virtual hosts. Same port.>> I think SPDY or like-protocols [...] give a
> crisp treatment to
> >> interactive, stateful workloads involving
> >>
> >> back-and-forth between client and server with multiplexing, fixing
> >> some problems with the hacks in HTTP-land from before.
> >
> > It sounds like at some level you're really talking about replacing the
> > built-in protocol with SPDY because SPDY is possibly a better baseline
> than
> > updating the existing protocol.  That's an interesting idea, I think this
> > project could evolve in that direction if there's demand for it.
>
> If only so there is a smaller set of arbitrary decisions to make about
> how to delimit messages...but if SPDY doesn't get widely deployed, or
> exacts an unacceptable performance penalty, it is game over.
>

Well, in our case HTTP is a clear win (but not replacement) and SPDY a
potential one (even as a replacement).  Even if SPDY is not widely adopted
it could still replace FEBE if there's a clear advantage to using it, I
don't know enough to make the call right now.


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 15:17  Dobes Vandermeer <[email protected]>
  parent: Merlin Moncure <[email protected]>
  1 sibling, 0 replies; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-30 15:17 UTC (permalink / raw)
  To: Merlin Moncure <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 9:57 PM, Merlin Moncure <[email protected]> wrote:

> On Thu, Mar 29, 2012 at 10:04 AM, Andrew Dunstan <[email protected]>
> wrote:
> > 1. I've been in discussion with some people about adding simple JSON
> extract
> > functions. We already have some (i.e. xpath()) for XML.
>
> Your json work is a great start in marrying document level database
> features with a relational backend.  My take is that storing rich data
> inside the database in json format, while tempting, is generally a
> mistake.  Unless the document is black box it should be decomposed and
> stored relationally and marked back up into a document as it goes out
> the door.
>

I found storing documents (with no schema) in CouchDB very sexy.  For
analytics purposes it's very handy to throw whatever data you've got into a
document and save it to the database and sort out the schema later, when
you have a need for it.  It could be you want to allow hundreds of
different fields in the document, and the document has a hierarchal
structure, and so on ... and yet there's no interest in spending time
figuring out how to map all that effectively into a SQL schema.  Some
things might start out a black box but become interesting for reporting
purposes later.

While I was doing this I always thought this would have been a better
approach for my previous project, an accounting application.  If I could
just have stored entities like invoice & customer as a single document that
is inserted, updated, etc. atomically it would be a lot simpler and faster
than having to break things out into columns and rows spread over various
tables.  Some fields are just "information" that isn't searched for or used
relationally but ends up getting a field (or a whole bunch of fields, like
your address) in the DB schema anyway.  Adding, removing, and changing
columns is always scary to me in the SQL database - when there's no schema
you can fix these things as part of the definition of your "view" instead.

So I kind of think the document database kind of bridges the gap between an
OODBMS and the RDBMS because the document is like a little cluster of
objects and then you populate your relations by mapping a view of those
documents and use that for analysis and search.

What I'm saying is that jsonpath probably isn't the whole story:
> another way of bulk moving json into native backend structures without
> parsing would also be very helpful.  For example, being able to cast a
> json document into a record or a record array would be just amazing.
>

Hmm whatever that exactly is, it does sound cool!


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 15:41  Robert Haas <[email protected]>
  parent: Dobes Vandermeer <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Robert Haas @ 2012-03-30 15:41 UTC (permalink / raw)
  To: Dobes Vandermeer <[email protected]>; +Cc: Daniel Farina <[email protected]>; Andrew Dunstan <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 10:55 AM, Dobes Vandermeer <[email protected]> wrote:
> Well, in our case HTTP is a clear win (but not replacement) and SPDY a
> potential one (even as a replacement).  Even if SPDY is not widely adopted
> it could still replace FEBE if there's a clear advantage to using it, I
> don't know enough to make the call right now.

I can see that there are some advantages to having an HTTP interface
to the database, but I think throwing our existing protocol out the
window or relegating it to the status of a second-class citizen would
be foolish.  HTTP is a non-trivial protocol that tends to impose lots
of escaping and de-escaping overhead which is unnecessary for people
who just want to connect to the database and run queries.  I can
completely understand that someone might want the ability to do GET
/db/table/pk and have that return an answer very, very quickly, by
bypassing the usual parser and planner and just firing off an
index-scan and returning the results as JSON or somesuch.  But I think
it would be a serious mistake to assume that GET /q?q=myquery is going
to come out better than what we have now in the general case.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 16:11  Andrew Dunstan <[email protected]>
  parent: Robert Haas <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Andrew Dunstan @ 2012-03-30 16:11 UTC (permalink / raw)
  To: Robert Haas <[email protected]>; +Cc: Dobes Vandermeer <[email protected]>; Daniel Farina <[email protected]>; pgsql-hackers



On 03/30/2012 11:41 AM, Robert Haas wrote:
> On Fri, Mar 30, 2012 at 10:55 AM, Dobes Vandermeer<[email protected]>  wrote:
>> Well, in our case HTTP is a clear win (but not replacement) and SPDY a
>> potential one (even as a replacement).  Even if SPDY is not widely adopted
>> it could still replace FEBE if there's a clear advantage to using it, I
>> don't know enough to make the call right now.
> I can see that there are some advantages to having an HTTP interface
> to the database, but I think throwing our existing protocol out the
> window or relegating it to the status of a second-class citizen would
> be foolish.


Right, I can't imagine it happening. And I wouldn't really be keen to 
add an alternative protocol either.

I could imagine a client which presented a SPDY interface to the world 
and translated it into standard calls, possibly via libpq.

It's well to remember that we are not a green fields project here.


> HTTP is a non-trivial protocol that tends to impose lots
> of escaping and de-escaping overhead which is unnecessary for people
> who just want to connect to the database and run queries.  I can
> completely understand that someone might want the ability to do GET
> /db/table/pk and have that return an answer very, very quickly, by
> bypassing the usual parser and planner and just firing off an
> index-scan and returning the results as JSON or somesuch.  But I think
> it would be a serious mistake to assume that GET /q?q=myquery is going
> to come out better than what we have now in the general case.


Indeed.

cheers

andrew




^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 17:21  Daniel Farina <[email protected]>
  parent: Andrew Dunstan <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Daniel Farina @ 2012-03-30 17:21 UTC (permalink / raw)
  To: Andrew Dunstan <[email protected]>; +Cc: Robert Haas <[email protected]>; Dobes Vandermeer <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 9:11 AM, Andrew Dunstan <[email protected]> wrote:
>
>
> On 03/30/2012 11:41 AM, Robert Haas wrote:
>>
>> On Fri, Mar 30, 2012 at 10:55 AM, Dobes Vandermeer<[email protected]>
>>  wrote:
>>>
>>> Well, in our case HTTP is a clear win (but not replacement) and SPDY a
>>> potential one (even as a replacement).  Even if SPDY is not widely
>>> adopted
>>> it could still replace FEBE if there's a clear advantage to using it, I
>>> don't know enough to make the call right now.
>>
>> I can see that there are some advantages to having an HTTP interface
>> to the database, but I think throwing our existing protocol out the
>> window or relegating it to the status of a second-class citizen would
>> be foolish.
>
>
>
> Right, I can't imagine it happening. And I wouldn't really be keen to add an
> alternative protocol either.
>
> I could imagine a client which presented a SPDY interface to the world and
> translated it into standard calls, possibly via libpq.

Any enhancement here that can't be used with libpq via, say, drop-in
.so seems unworkable to me, and that's why any solution that is
basically proxying to the database is basically a non-starter outside
the very earliest prototyping stages.  The tuple scanning and protocol
semantics can and even should remain the same, especially at first.

What I really want is for a mechanism to push FEBE messages down into
a transport in a extendable way (not unlike SSL, except this time
sensitive to FEBE message boundaries), taking unmodified libpq
binaries (but with a hook, of course...) into consideration.

> It's well to remember that we are not a green fields project here.

Perhaps not, which is a good reason to not put cart before horse --
I've tried to be very careful to suggest that real useful
caching/proxy/routing software needs to appear that can work,
unmodified, with hypothetical FEBE-over-SPDY communications, and
accomplish some interesting use cases, and also very careful to
suggest this is a *transport*-level feature.  It think would be fair
to demand: "Okay, when Apache or NGINX or a credible embedded web
server in some language can accomplish something useful for us if we
absorb this complexity, let's put this on the table".

To have to write continuously write poolers, cachers and redirectors
(for any reason: HA, partitioning, or even just clean-looking domain
names...) is a burden on community resources. The criteria for success
is to expand the number of projects attacking these problems and
decrease the cost of writing one, and I think demanding a straw-man
implementation of, say, a query cacher or pooler and seeing how that
implementation feels is not unreasonable along with accepting the
burden of maintenance.

Robert Haas Wrote:
>> HTTP is a non-trivial protocol that tends to impose lots
>> of escaping and de-escaping overhead which is unnecessary for people
>> who just want to connect to the database and run queries.  I can
>> completely understand that someone might want the ability to do GET
>> /db/table/pk and have that return an answer very, very quickly, by
>> bypassing the usual parser and planner and just firing off an
>> index-scan and returning the results as JSON or somesuch.  But I think
>> it would be a serious mistake to assume that GET /q?q=myquery is going
>> to come out better than what we have now in the general case.

I don't think this holds for SPDY, which is the only way I even
entertain thoughts of doing this.  I want FEBE, at least at first and
probably even then for quite a long while, just with framing and
headers (during start-up) that other software will understand.

Control frames are like this:

+----------------------------------+
|C| Version(15bits) | Type(16bits) |
+----------------------------------+
| Flags (8)  |  Length (24 bits)   |
+----------------------------------+
|               Data               |
+----------------------------------+

Data frames are like this:

+----------------------------------+
|C|       Stream-ID (31bits)       |
+----------------------------------+
| Flags (8)  |  Length (24 bits)   |
+----------------------------------+
|               Data               |
+----------------------------------+

The (XML, not HTML, I read this with "view source", or try "curl") RFC
draft is here:

http://mbelshe.github.com/SPDY-Specification/draft-mbelshe-spdy-00.xml


-- 
fdr



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-30 17:44  Daniel Farina <[email protected]>
  parent: Daniel Farina <[email protected]>
  0 siblings, 1 reply; 8673+ messages in thread

From: Daniel Farina @ 2012-03-30 17:44 UTC (permalink / raw)
  To: Andrew Dunstan <[email protected]>; +Cc: Robert Haas <[email protected]>; Dobes Vandermeer <[email protected]>; pgsql-hackers

On Fri, Mar 30, 2012 at 10:21 AM, Daniel Farina <[email protected]> wrote:
> Any enhancement here that can't be used with libpq via, say, drop-in
> .so seems unworkable to me, and that's why any solution that is
> basically proxying to the database is basically a non-starter outside
> the very earliest prototyping stages.  The tuple scanning and protocol
> semantics can and even should remain the same, especially at first.

I should add: proxying could work well if libpq had all the right
hooks. The server could remain ignorant.  Regardless, upstream changes
result.

-- 
fdr



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* Re: HTTP Frontend? (and a brief thought on materialized views)
@ 2012-03-31 13:37  Dobes Vandermeer <[email protected]>
  parent: Daniel Farina <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Dobes Vandermeer @ 2012-03-31 13:37 UTC (permalink / raw)
  To: Daniel Farina <[email protected]>; +Cc: Andrew Dunstan <[email protected]>; Robert Haas <[email protected]>; pgsql-hackers

On Sat, Mar 31, 2012 at 1:44 AM, Daniel Farina <[email protected]> wrote:

> On Fri, Mar 30, 2012 at 10:21 AM, Daniel Farina <[email protected]> wrote:
> > Any enhancement here that can't be used with libpq via, say, drop-in
> > .so seems unworkable to me, and that's why any solution that is
> > basically proxying to the database is basically a non-starter outside
> > the very earliest prototyping stages.  The tuple scanning and protocol
> > semantics can and even should remain the same, especially at first.
>
> I should add: proxying could work well if libpq had all the right
> hooks. The server could remain ignorant.  Regardless, upstream changes
> result.
>

Just to be clear, what you are saying that writing a process that accepts
requests by HTTP and translates them into requests using the existing
protocol to send to the server would have unacceptable performance?  Or is
there something else about it that is a non-starter?


^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v52 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--gp2pyozrd5pweboh
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v52-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v52 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--gp2pyozrd5pweboh
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v52-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v52 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--gp2pyozrd5pweboh
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v52-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v52 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--gp2pyozrd5pweboh
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v52-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v52 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--gp2pyozrd5pweboh
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v52-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v52 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--gp2pyozrd5pweboh
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v52-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v52 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--gp2pyozrd5pweboh
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v52-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v51 08/10] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--r2slln3zpmilwu22
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v51-0009-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH v50 7/8] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..df092dc999a 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool		accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index c85166ba849..ff34e246469 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -237,6 +238,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index b4269a3b102..2e3926e1d13 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1465,7 +1466,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid			dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index a9092fc2382..d553bd5dbff 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1760,9 +1761,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid			dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..92ed34128e5 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+				continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--brnevsqjnuzpyok4
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v50-0008-Reserve-replication-slots-specifically-for-REPAC.patch"



^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread

* [PATCH] Introduce an option to make logical replication database specific.
@ 2026-04-03 10:34  Antonin Houska <[email protected]>
  0 siblings, 0 replies; 8673+ messages in thread

From: Antonin Houska @ 2026-04-03 10:34 UTC (permalink / raw)

By default, the logical decoding assumes access to shared catalogs, so the
snapshot builder needs to consider cluster-wide XIDs during startup. That in
turn means that, if any transaction is already running (and has XID assigned),
the snapshot builder needs to wait for its completion, as it does not know if
that transaction performed catalog changes earlier.

Possible problem with this concept is that if REPACK (CONCURRENTLY) is running
in some database, backends running the same command in other databases get
stuck until the first one has committed. Thus only as single backend in the
cluster can run REPACK (CONCURRENTLY) at any time. Likewise, REPACK
(CONCURRENTLY) can block walsenders starting on behalf of subscriptions
throughout the cluster.

This patch introduces the possibility for a backend to declare that its output
plugin does not use shared catalogs (i.e. catalogs that can be changed by
transactions running in other databases in the cluster). In that case, no
snapshot the backend will use during the decoding needs to contain information
about transactions running in other databases. Thus the snapshot builder only
needs to wait for completion of transactions in the current database.

Currently we only use this option in the REPACK background worker. It could
possibly be used in walsender involved in logical replication too, however
that would need thorough analysis of its output plugin.

The patch bumps WAL version number, due to a new field in xl_running_xacts.
---
 contrib/pg_visibility/pg_visibility.c       |  4 ++--
 src/backend/access/index/genam.c            | 18 ++++++++++++++++
 src/backend/access/rmgrdesc/standbydesc.c   |  2 ++
 src/backend/access/transam/xlog.c           |  2 +-
 src/backend/access/transam/xlogfuncs.c      |  2 +-
 src/backend/commands/repack_worker.c        |  8 +++++++
 src/backend/postmaster/bgwriter.c           |  2 +-
 src/backend/replication/logical/snapbuild.c | 10 ++++++++-
 src/backend/replication/slot.c              | 12 +++++++++--
 src/backend/storage/ipc/procarray.c         | 23 ++++++++++++++++++++-
 src/backend/storage/ipc/standby.c           | 14 +++++++++++--
 src/include/access/genam.h                  |  8 +++++++
 src/include/access/xlog_internal.h          |  2 +-
 src/include/storage/procarray.h             |  2 +-
 src/include/storage/standby.h               |  3 ++-
 src/include/storage/standbydefs.h           |  1 +
 16 files changed, 99 insertions(+), 14 deletions(-)

diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c
index dfab0b64cf5..d564bd2a00c 100644
--- a/contrib/pg_visibility/pg_visibility.c
+++ b/contrib/pg_visibility/pg_visibility.c
@@ -621,7 +621,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 	else if (rel == NULL || rel->rd_rel->relisshared)
 	{
 		/* Shared relation: take into account all running xids */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestRunningXid;
@@ -632,7 +632,7 @@ GetStrictOldestNonRemovableTransactionId(Relation rel)
 		 * Normal relation: take into account xids running within the current
 		 * database
 		 */
-		runningTransactions = GetRunningTransactionData();
+		runningTransactions = GetRunningTransactionData(InvalidOid);
 		LWLockRelease(ProcArrayLock);
 		LWLockRelease(XidGenLock);
 		return runningTransactions->oldestDatabaseRunningXid;
diff --git a/src/backend/access/index/genam.c b/src/backend/access/index/genam.c
index 1408989c568..97ea882cf71 100644
--- a/src/backend/access/index/genam.c
+++ b/src/backend/access/index/genam.c
@@ -37,6 +37,14 @@
 #include "utils/ruleutils.h"
 #include "utils/snapmgr.h"
 
+/*
+ * If a backend is going to do logical decoding and if the output plugin does
+ * not need to access shared catalogs, setting this variable to false can make
+ * the decoding startup faster. In particular, the backend will not need to
+ * wait for completion of already running transactions in other databases.
+ */
+bool accessSharedCatalogsInDecoding = true;
+
 
 /* ----------------------------------------------------------------
  *		general access method routines
@@ -394,6 +402,16 @@ systable_beginscan(Relation heapRelation,
 	SysScanDesc sysscan;
 	Relation	irel;
 
+	/*
+	 * If this backend promised that it won't access shared catalogs during
+	 * logical decoding, this seems to be the right place to check.
+	 *
+	 * XXX Should this be ereport(ERROR) ?
+	 */
+	Assert(!HistoricSnapshotActive() ||
+		   accessSharedCatalogsInDecoding ||
+		   !heapRelation->rd_rel->relisshared);
+
 	if (indexOK &&
 		!IgnoreSystemIndexes &&
 		!ReindexIsProcessingIndex(indexId))
diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c
index 0a291354ae2..685d1bdb024 100644
--- a/src/backend/access/rmgrdesc/standbydesc.c
+++ b/src/backend/access/rmgrdesc/standbydesc.c
@@ -41,6 +41,8 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec)
 		for (i = 0; i < xlrec->subxcnt; i++)
 			appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]);
 	}
+
+	appendStringInfo(buf, "; dbid: %u", xlrec->dbid);
 }
 
 void
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2c1c6f88b74..8da8eff93e4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7336,7 +7336,7 @@ CreateCheckPoint(int flags)
 	 * recovery we don't need to write running xact data.
 	 */
 	if (!shutdown && XLogStandbyInfoActive())
-		LogStandbySnapshot();
+		LogStandbySnapshot(InvalidOid);
 
 	START_CRIT_SECTION();
 
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 65bbaeda59c..0f5979691e6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -245,7 +245,7 @@ pg_log_standby_snapshot(PG_FUNCTION_ARGS)
 				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
 				 errmsg("pg_log_standby_snapshot() can only be used if \"wal_level\" >= \"replica\"")));
 
-	recptr = LogStandbySnapshot();
+	recptr = LogStandbySnapshot(InvalidOid);
 
 	/*
 	 * As a convenience, return the WAL location of the last inserted record
diff --git a/src/backend/commands/repack_worker.c b/src/backend/commands/repack_worker.c
index 00b21ede481..c25dbeadff3 100644
--- a/src/backend/commands/repack_worker.c
+++ b/src/backend/commands/repack_worker.c
@@ -16,6 +16,7 @@
  */
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/table.h"
 #include "access/xlog_internal.h"
 #include "access/xlogutils.h"
@@ -233,6 +234,13 @@ repack_setup_logical_decoding(Oid relid)
 
 	EnsureLogicalDecodingEnabled();
 
+	/*
+	 * By declaring that our output plugin does not need shared catalogs, we
+	 * avoid waiting for completion of transactions running in other databases
+	 * than the one we're connected to.
+	 */
+	accessSharedCatalogsInDecoding = false;
+
 	/*
 	 * Neither prepare_write nor do_write callback nor update_progress is
 	 * useful for us.
diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c
index 1d8947774a9..a30de4262eb 100644
--- a/src/backend/postmaster/bgwriter.c
+++ b/src/backend/postmaster/bgwriter.c
@@ -289,7 +289,7 @@ BackgroundWriterMain(const void *startup_data, size_t startup_data_len)
 			if (now >= timeout &&
 				last_snapshot_lsn <= GetLastImportantRecPtr())
 			{
-				last_snapshot_lsn = LogStandbySnapshot();
+				last_snapshot_lsn = LogStandbySnapshot(InvalidOid);
 				last_snapshot_ts = now;
 			}
 		}
diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 0f6b5df322c..16009fa75a4 100644
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -125,6 +125,7 @@
 #include <sys/stat.h>
 #include <unistd.h>
 
+#include "access/genam.h"
 #include "access/heapam_xlog.h"
 #include "access/transam.h"
 #include "access/xact.h"
@@ -1470,7 +1471,14 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff)
 	 */
 	if (!RecoveryInProgress())
 	{
-		LogStandbySnapshot();
+		Oid		dbid;
+
+		/*
+		 * Only consider transactions of the current database if our plugin is
+		 * not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		LogStandbySnapshot(dbid);
 	}
 }
 
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index fe6bfeba25e..0d415f5f8f2 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -39,6 +39,7 @@
 #include <unistd.h>
 #include <sys/stat.h>
 
+#include "access/genam.h"
 #include "access/transam.h"
 #include "access/xlog_internal.h"
 #include "access/xlogrecovery.h"
@@ -1776,9 +1777,16 @@ ReplicationSlotReserveWal(void)
 	if (!RecoveryInProgress() && SlotIsLogical(slot))
 	{
 		XLogRecPtr	flushptr;
+		Oid		dbid;
 
-		/* make sure we have enough information to start */
-		flushptr = LogStandbySnapshot();
+		/*
+		 * Make sure we have enough information to start.
+		 *
+		 * Only consider transactions of the current database if our plugin
+		 * is not supposed to access shared catalogs.
+		 */
+		dbid = accessSharedCatalogsInDecoding ? InvalidOid : MyDatabaseId;
+		flushptr = LogStandbySnapshot(dbid);
 
 		/* and make sure it's fsynced to disk */
 		XLogFlush(flushptr);
diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c
index cc207cb56e3..49d1bf4bfac 100644
--- a/src/backend/storage/ipc/procarray.c
+++ b/src/backend/storage/ipc/procarray.c
@@ -2631,9 +2631,11 @@ ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc)
  *
  * Note that if any transaction has overflowed its cached subtransactions
  * then there is no real need include any subtransactions.
+ *
+ * If 'dbid' is valid, only gather transactions running in that database.
  */
 RunningTransactions
-GetRunningTransactionData(void)
+GetRunningTransactionData(Oid dbid)
 {
 	/* result workspace */
 	static RunningTransactionsData CurrentRunningXactsData;
@@ -2708,6 +2710,18 @@ GetRunningTransactionData(void)
 		if (!TransactionIdIsValid(xid))
 			continue;
 
+		/*
+		 * Filter by database OID if requested.
+		 */
+		if (OidIsValid(dbid))
+		{
+			int			pgprocno = arrayP->pgprocnos[index];
+			PGPROC	   *proc = &allProcs[pgprocno];
+
+			if (proc->databaseId != dbid)
+				continue;
+		}
+
 		/*
 		 * Be careful not to exclude any xids before calculating the values of
 		 * oldestRunningXid and suboverflowed, since these are used to clean
@@ -2758,6 +2772,12 @@ GetRunningTransactionData(void)
 			PGPROC	   *proc = &allProcs[pgprocno];
 			int			nsubxids;
 
+			/*
+			 * Filter by database OID if requested.
+			 */
+			if (OidIsValid(dbid) && proc->databaseId != dbid)
+					continue;
+
 			/*
 			 * Save subtransaction XIDs. Other backends can't add or remove
 			 * entries while we're holding XidGenLock.
@@ -2791,6 +2811,7 @@ GetRunningTransactionData(void)
 	 * increases if slots do.
 	 */
 
+	CurrentRunningXacts->dbid = dbid;
 	CurrentRunningXacts->xcnt = count - subcount;
 	CurrentRunningXacts->subxcnt = subcount;
 	CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY;
diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c
index de9092fdf5b..c653ea742bc 100644
--- a/src/backend/storage/ipc/standby.c
+++ b/src/backend/storage/ipc/standby.c
@@ -1188,6 +1188,14 @@ standby_redo(XLogReaderState *record)
 		xl_running_xacts *xlrec = (xl_running_xacts *) XLogRecGetData(record);
 		RunningTransactionsData running;
 
+		/*
+		 * Records issued for specific database are not suitable for physical
+		 * replication because that affects the whole cluster. In particular,
+		 * the list of XID is probably incomplete here.
+		 */
+		if (OidIsValid(xlrec->dbid))
+			return;
+
 		running.xcnt = xlrec->xcnt;
 		running.subxcnt = xlrec->subxcnt;
 		running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY;
@@ -1277,11 +1285,12 @@ standby_redo(XLogReaderState *record)
  * as there's no independent knob to just enable logical decoding. For
  * details of how this is used, check snapbuild.c's introductory comment.
  *
+ * If 'dbid' is valid, only gather transactions running in that database.
  *
  * Returns the RecPtr of the last inserted record.
  */
 XLogRecPtr
-LogStandbySnapshot(void)
+LogStandbySnapshot(Oid dbid)
 {
 	XLogRecPtr	recptr;
 	RunningTransactions running;
@@ -1314,7 +1323,7 @@ LogStandbySnapshot(void)
 	 * Log details of all in-progress transactions. This should be the last
 	 * record we write, because standby will open up when it sees this.
 	 */
-	running = GetRunningTransactionData();
+	running = GetRunningTransactionData(dbid);
 
 	/*
 	 * GetRunningTransactionData() acquired ProcArrayLock, we must release it.
@@ -1358,6 +1367,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts)
 	xl_running_xacts xlrec;
 	XLogRecPtr	recptr;
 
+	xlrec.dbid = CurrRunningXacts->dbid;
 	xlrec.xcnt = CurrRunningXacts->xcnt;
 	xlrec.subxcnt = CurrRunningXacts->subxcnt;
 	xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY);
diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index b69320a7fc8..56c573399ab 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -136,6 +136,14 @@ typedef struct IndexOrderByDistance
 	bool		isnull;
 } IndexOrderByDistance;
 
+/*
+ * Is the backend interested in shared catalogs when performing logical
+ * decoding?
+ *
+ * XXX Is there a better place for this declaration?
+ */
+extern bool accessSharedCatalogsInDecoding;
+
 /*
  * generalized index_ interface routines (in indexam.c)
  */
diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h
index 755835d63bf..ae19982d88d 100644
--- a/src/include/access/xlog_internal.h
+++ b/src/include/access/xlog_internal.h
@@ -31,7 +31,7 @@
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD11E	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD11F	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {
diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h
index abdf021e66e..377b3060b9f 100644
--- a/src/include/storage/procarray.h
+++ b/src/include/storage/procarray.h
@@ -49,7 +49,7 @@ extern bool ProcArrayInstallImportedXmin(TransactionId xmin,
 										 VirtualTransactionId *sourcevxid);
 extern bool ProcArrayInstallRestoredXmin(TransactionId xmin, PGPROC *proc);
 
-extern RunningTransactions GetRunningTransactionData(void);
+extern RunningTransactions GetRunningTransactionData(Oid dbid);
 
 extern bool TransactionIdIsInProgress(TransactionId xid);
 extern TransactionId GetOldestNonRemovableTransactionId(Relation rel);
diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h
index 6a314c693cd..8715c08e94f 100644
--- a/src/include/storage/standby.h
+++ b/src/include/storage/standby.h
@@ -126,6 +126,7 @@ typedef enum
 
 typedef struct RunningTransactionsData
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	subxids_array_status subxid_status;
@@ -143,7 +144,7 @@ typedef RunningTransactionsData *RunningTransactions;
 extern void LogAccessExclusiveLock(Oid dbOid, Oid relOid);
 extern void LogAccessExclusiveLockPrepare(void);
 
-extern XLogRecPtr LogStandbySnapshot(void);
+extern XLogRecPtr LogStandbySnapshot(Oid dbid);
 extern void LogStandbyInvalidations(int nmsgs, SharedInvalidationMessage *msgs,
 									bool relcacheInitFileInval);
 
diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h
index 231d251fd51..e75b7078766 100644
--- a/src/include/storage/standbydefs.h
+++ b/src/include/storage/standbydefs.h
@@ -46,6 +46,7 @@ typedef struct xl_standby_locks
  */
 typedef struct xl_running_xacts
 {
+	Oid			dbid;			/* only track xacts in this database */
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
 	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
-- 
2.47.3


--=-=-=--





^ permalink  raw  reply  [nested|flat] 8673+ messages in thread