Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-27 Thread Albe Laurenz
Chris Travers wrote: > If it is truly idle in a transaction, maybe it has locks that are holding up other transactions? > > Locks are usually held until commit time, except advisory locks iirc but those have to be explicitly > checked, so if you don't know if you are using them you probably aren't

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-27 Thread Chris Travers
If it is truly idle in a transaction, maybe it has locks that are holding up other transactions? Locks are usually held until commit time, except advisory locks iirc but those have to be explicitly checked, so if you don't know if you are using them you probably aren't. Long-running transactions

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-27 Thread Albe Laurenz
Scot Kreienkamp wrote: >> You could try to "strace" the postmaster during a connection attempt >> and see what happens. Maybe that helps to spot the place where >> things go wrong. > [Scot Kreienkamp] > I'm willing to give it a try, but I've never done it before. What do I need to do? "man stra

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-26 Thread Scot Kreienkamp
> -Original Message- > From: pgsql-general-ow...@postgresql.org [mailto:pgsql-general- > ow...@postgresql.org] On Behalf Of Albe Laurenz > Sent: Wednesday, September 26, 2012 5:15 AM > To: Scot Kreienkamp; pgsql-general@postgresql.org > Subject: Re: [GENERAL] idle in

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-26 Thread Scot Kreienkamp
> -Original Message- > From: pgsql-general-ow...@postgresql.org [mailto:pgsql-general- > ow...@postgresql.org] On Behalf Of Thomas Kellerer > Sent: Tuesday, September 25, 2012 5:25 PM > To: pgsql-general@postgresql.org > Subject: Re: [GENERAL] idle in transaction

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-26 Thread Albe Laurenz
Scot Kreienkamp wrote: > I have a problem that I've been struggling with for quite some time. Every once in a while I will get > a connection that goes to idle in transaction on an in-house programmed application that connects with > JDBC. That happens fairly regularly and the programmers are tryi

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-25 Thread Thomas Kellerer
Scot Kreienkamp wrote on 25.09.2012 22:35: The application is using a pooler and generally runs around 100 connections, but I've seen it as high as 200 during the day for normal use. It's on a large server; 64 cores total and about 500 gigs of memory. That's one of the reasons I left it at 512

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-25 Thread John R Pierce
On 09/25/12 1:35 PM, Scot Kreienkamp wrote: The problem is how do I investigate this when PG is entirely unresponsive? Why is it becoming unresponsive, and how do I prevent the PG server from becoming unresponsive? I think I'd push that 9.1.latest upgrade ASAP, and then see if this problem

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-25 Thread Scot Kreienkamp
> -Original Message- > From: pgsql-general-ow...@postgresql.org [mailto:pgsql-general- > ow...@postgresql.org] On Behalf Of John R Pierce > Sent: Tuesday, September 25, 2012 3:53 PM > To: pgsql-general@postgresql.org > Subject: Re: [GENERAL] idle in transaction

Re: [GENERAL] idle in transaction query makes server unresponsive

2012-09-25 Thread John R Pierce
On 09/25/12 12:23 PM, Scot Kreienkamp wrote: I have a problem that I've been struggling with for quite some time. Every once in a while I will get a connection that goes to idle in transaction on an in-house programmed application that connects with JDBC. That happens fairly regularly and t

[GENERAL] idle in transaction query makes server unresponsive

2012-09-25 Thread Scot Kreienkamp
Hi everyone, I have a problem that I've been struggling with for quite some time. Every once in a while I will get a connection that goes to idle in transaction on an in-house programmed application that connects with JDBC. That happens fairly regularly and the programmers are trying to clean