Re: BUG: endless lseek(.., SEEK_END) from select queries on x64 builds

From: Stephen Frost <sfrost(at)snowman(dot)net>
To: Markus Schulz <msc(at)antzsystem(dot)de>
Cc: pgsql-performance(at)postgresql(dot)org
Subject: Re: BUG: endless lseek(.., SEEK_END) from select queries on x64 builds
Date: 2013-02-22 13:37:23
Message-ID: 20130222133723.GR16142@tamriel.snowman.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Markus,

* Markus Schulz (msc(at)antzsystem(dot)de) wrote:
> as you can see the query runs fine.
> I can run this query from a bash-psql-while-loop/jdbc-cli-tool
> endless without any problems.
> so far so good.
[...]
> JBoss EAP 5.1.2 with connection pooling and xa-datasource/two-phase-commits
> (transactions on multiple datasources needed)
> *and* *<prepared-statement-cache-size>1000</prepared-statement-cache-size>*
>
> i can run the query four times with good performance and after that postgresql
> starts with the strange lseek() behavior.

It sounds like your bash script and JBoss are doing something different.
Would it be possible for you to turn on log_statements = 'all' in PG,
see what's different, and then update the bash/psql script to do exactly
what JBoss does, and see if you can reproduce it that way?

It certainly looks like a PG bug, but it'd be a lot easier to debug with
a simple, well-defined test case which shows the failure.

Thanks!

Stephen

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message jackrg 2013-02-22 16:36:16 Avoiding Recheck Cond when using Select Distinct
Previous Message Heikki Linnakangas 2013-02-22 13:35:25 Re: BUG: endless lseek(.., SEEK_END) from select queries on x64 builds