pg.ddx.io pgsql-interfaces@postgresql.org mailing list archive
help / color / mirror / Atom feed8k limit
6+ messages / 6 participants
[nested] [flat]
* 8k limit
@ 1999-10-26 12:44 StOo <stuart@comodo.net>
0 siblings, 1 reply; 6+ messages in thread
From: StOo @ 1999-10-26 12:44 UTC (permalink / raw)
To: pgsql-interfaces
Hello,
Im trying to store a string of ascii characters greater than 8k. Is there a
tunable which will allow me to alter the 8k max limit? Im pretty sure this has
been discussed here before =). Next question is whats the max number of large
objects you can store? I think its about 4 billion , am i wrong?
+-=-=-=-=-=-=-=StOo=-=-=-=-=-=-=-+
Stuart Hodgkinson Software Engineer
stuart@comodo.net
^ permalink raw reply [nested|flat] 6+ messages in thread
* RE: [INTERFACES] 8k limit
@ 1999-10-26 13:31 Ansley, Michael <Michael.Ansley@intec.co.za>
0 siblings, 1 reply; 6+ messages in thread
From: Ansley, Michael @ 1999-10-26 13:31 UTC (permalink / raw)
To: 'stuart@comodo.net' <stuart@comodo.net>; pgsql-interfaces
The 8kB limit you are refering to is probably the tuple-length limit. This
can be altered to a maximum of 64kB (I think, might only be 32kB), but you
need to recompile your server, and initdb. This limit will not be removed
until at least version 7.
There is also a limit on the length of a query string, if you are using a
6.5.x production release. This limit is set to twice the tuple-length
limit, so if your block-size (tuple-length) is set to 8kB, then your maximum
query string length will be 16kB. If you recompile your server to up the
block-size to, say, 32kB, then your query string limit will automatically be
altered to 64kB.
As far as the number of los go, I'm afraid I can't help you on that one.
MikeA
>> -----Original Message-----
>> From: StOo [mailto:stuart@comodo.net]
>> Sent: Tuesday, October 26, 1999 2:45 PM
>> To: pgsql-interfaces@postgreSQL.org
>> Subject: [INTERFACES] 8k limit
>>
>>
>>
>> Hello,
>> Im trying to store a string of ascii characters greater
>> than 8k. Is there a
>> tunable which will allow me to alter the 8k max limit? Im
>> pretty sure this has
>> been discussed here before =). Next question is whats the
>> max number of large
>> objects you can store? I think its about 4 billion , am i wrong?
>>
>>
>>
>> +-=-=-=-=-=-=-=StOo=-=-=-=-=-=-=-+
>>
>> Stuart Hodgkinson Software Engineer
>> stuart@comodo.net
>>
>>
>>
>> ************
>>
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: [INTERFACES] 8k limit
@ 1999-10-26 16:16 Tom Lane <tgl@sss.pgh.pa.us>
parent: Ansley, Michael <Michael.Ansley@intec.co.za>
0 siblings, 1 reply; 6+ messages in thread
From: Tom Lane @ 1999-10-26 16:16 UTC (permalink / raw)
To: Ansley, Michael <Michael.Ansley@intec.co.za>; +Cc: 'stuart@comodo.net' <stuart@comodo.net>; pgsql-interfaces
"Ansley, Michael" <Michael.Ansley@intec.co.za> writes:
> The 8kB limit you are refering to is probably the tuple-length limit. This
> can be altered to a maximum of 64kB (I think, might only be 32kB),
32k; otherwise Michael's comments are accurate.
>>> been discussed here before =). Next question is whats the
>>> max number of large
>>> objects you can store? I think its about 4 billion , am i wrong?
Each LO results in the creation of two files in your database directory,
which means performance will get pretty horrible when you get past a
few thousand LOs (at least on most flavors of Unix, which are not
optimized for huge directories).
There has been discussion of fixing this, eg by making a subdirectory
tree for LOs, but I'd rather see us put our effort into eliminating
the tuple length limit.
regards, tom lane
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: [INTERFACES] 8k limit
@ 1999-10-27 04:40 Gene Selkov, Jr. <selkovjr@mcs.anl.gov>
parent: StOo <stuart@comodo.net>
0 siblings, 1 reply; 6+ messages in thread
From: Gene Selkov, Jr. @ 1999-10-27 04:40 UTC (permalink / raw)
To: pgsql-interfaces; stuart@comodo.net
> Next question is whats the max number of large
> objects you can store? I think its about 4 billion , am i wrong?
Don't know the exact theortical limit, but I believe it depends on the
capability of your file system, rather than pointer size and such,
because each lo is a file.
For example, my file system can have 1064960 files in it (one can find
that out with fsck), but the practical constraint will be much
lower. I find it difficult to deal with 30000 files in one directory,
for example. The time it takes to open a file in such directory is
usually on the order of seconds.
--Gene
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: [INTERFACES] 8k limit
@ 1999-10-27 06:29 Tim Joyce <tim@hoop.co.uk>
parent: Gene Selkov, Jr. <selkovjr@mcs.anl.gov>
0 siblings, 0 replies; 6+ messages in thread
From: Tim Joyce @ 1999-10-27 06:29 UTC (permalink / raw)
To: pgsql-interfaces; stuart@comodo.net; Gene Selkov, Jr. <selkovjr@mcs.anl.gov>
>
> For example, my file system can have 1064960 files in it (one can find
> that out with fsck), but the practical constraint will be much
> lower. I find it difficult to deal with 30000 files in one directory,
> for example. The time it takes to open a file in such directory is
> usually on the order of seconds.
I don't know if this info is useful, but;
I have a directory with 150K files in it, and can open files in it without
noticable delay (this is not through postgres). Before setting up this dir,
i tried to find out the limit of how many files is sensible. I couldn't
find any info so just thought i'd try it, and it appears to work fine. If
you think I am going to hit problems (I am serving jpgs from this directory
for a web site), please let me know.
running on
Red Hat Linux release 5.2 (Apollo)
Kernel 2.0.36 on an i686
cheers
timj
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: [INTERFACES] 8k limit
@ 1999-10-27 06:41 marten@feki.toppoint.de
parent: Tom Lane <tgl@sss.pgh.pa.us>
0 siblings, 0 replies; 6+ messages in thread
From: marten@feki.toppoint.de @ 1999-10-27 06:41 UTC (permalink / raw)
To: tgl@sss.pgh.pa.us; +Cc: Michael.Ansley@intec.co.za; stuart@comodo.net; pgsql-interfaces
>
> There has been discussion of fixing this, eg by making a subdirectory
> tree for LOs, but I'd rather see us put our effort into eliminating
> the tuple length limit.
>
This would be very nice - because this also means to get SQL
statements that may really be 8k long !
Perhaps you all remember a rather complex view statement I posted here
some days ago, where postgresql said something like "rule too large" and
this seems to be a limitation for me. Actually as a user one does not know
when this limit is reached by the system.
All "we" (the users :-)) know about the 8192 byte limit of an SQL statement,
but this is not the whole truth.
Marten
^ permalink raw reply [nested|flat] 6+ messages in thread
end of thread, other threads:[~1999-10-27 06:41 UTC | newest]
Thread overview: 6+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
1999-10-26 12:44 8k limit StOo <stuart@comodo.net>
1999-10-27 04:40 ` Gene Selkov, Jr. <selkovjr@mcs.anl.gov>
1999-10-27 06:29 ` Tim Joyce <tim@hoop.co.uk>
1999-10-26 13:31 RE: [INTERFACES] 8k limit Ansley, Michael <Michael.Ansley@intec.co.za>
1999-10-26 16:16 ` Tom Lane <tgl@sss.pgh.pa.us>
1999-10-27 06:41 ` marten@feki.toppoint.de
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox