public inbox for [email protected]  

help / color / mirror / Atom feed

From: Bharath Rupireddy <[email protected]>
To: Dilip Kumar <[email protected]>
Cc: Andrey Borodin <[email protected]>
Cc: Laurenz Albe <[email protected]>
Cc: PostgreSQL Hackers <[email protected]>
Cc: SATYANARAYANA NARLAPURAM <[email protected]>
Subject: Re: An attempt to avoid locally-committed-but-not-replicated-to-standby-transactions in synchronous replication
Date: Tue, 10 May 2022 13:29:59 +0530
Message-ID: <CALj2ACVuj=Fo55NdWJy=fwweC-+pi7eQPRN3Kij3PQQcWFtYxQ@mail.gmail.com> (raw)
In-Reply-To: <CAFiTN-tzW+pFXGh-NqZCJmBGPsd2EHtvYvSxvsZTH7=cWsdPMw@mail.gmail.com>
References: <CALj2ACUrOB59QaE6=jF2cFAyv1MR7fzD8tr4YM5+OwEYG1SNzA@mail.gmail.com>
	<[email protected]>
	<CALj2ACWoB3k0kfjA7JxJgskVGGeE3jWzmGRPjP9QTRSCgSjhOg@mail.gmail.com>
	<[email protected]>
	<CAFiTN-tzW+pFXGh-NqZCJmBGPsd2EHtvYvSxvsZTH7=cWsdPMw@mail.gmail.com>

On Tue, May 10, 2022 at 1:18 PM Dilip Kumar <[email protected]> wrote:
>
> On Mon, May 9, 2022 at 4:39 PM Andrey Borodin <[email protected]> wrote:
>
> > > On 9 May 2022, at 14:44, Dilip Kumar <[email protected]> wrote:
> > >
> > > IMHO, making it wait for some amount of time, based on GUC is not a
> > > complete solution.  It is just a hack to avoid the problem in some
> > > cases.
> >
> > Disallowing cancelation of locally committed transactions is not a hack. It's removing of a hack that was erroneously installed to make backend responsible to Ctrl+C (or client side statement timeout).
>
> I might be missing something but based on my understanding the
> approach is not disallowing the query cancellation but it is just
> adding the configuration for how much to delay before canceling the
> query.  That's the reason I mentioned that this is not a guarenteed
> solution.  I mean with this configuration value also you can not avoid
> problems in all the cases, right?

Yes Dilip, the proposed GUC in v1 patch doesn't allow waiting forever
for sync repl ack, in other words, doesn't allow blocking the pending
query cancels or proc die interrupts forever. The backends may linger
in case repl ack isn't received or sync replicas aren't reachable?
Users may have to set the GUC to a 'reasonable value'.

If okay, I can make the GUC behave this way - value 0 existing
behaviour i.e. no wait for sync repl ack, just process query cancels
and proc die interrupts immediately; value -1 wait unboundedly for the
ack; value > 0 wait for specified milliseconds for the ack.

Regards,
Bharath Rupireddy.

view thread (37+ messages)  latest in thread

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients using the --to and --cc options:
  reply via email

  To: [email protected]
  Cc: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]
  Subject: Re: An attempt to avoid locally-committed-but-not-replicated-to-standby-transactions in synchronous replication
  In-Reply-To: <CALj2ACVuj=Fo55NdWJy=fwweC-+pi7eQPRN3Kij3PQQcWFtYxQ@mail.gmail.com>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox