pg.ddx.io  pgsql-admin@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: Konstantin Pelepelin <checat@pochtamt.ru>
To: mike@thegodshalls.com
Cc: pgsql-admin@postgresql.org
Subject: Re: are there ways for 'idle timeout'?
Date: Mon, 12 Jul 2004 13:26:41 +0400
Message-ID: <200407121326.41770.checat@pochtamt.ru> (raw)
In-Reply-To: <1089430151.10763.17.camel@localhost.localdomain>
References: <opsavj6izuhocrah@cathost.of.dtf.ru>
	<1089430151.10763.17.camel@localhost.localdomain>

Hi!

> The best solution would be to fix the web app but...
...but it's not a problem in my app, but in development 
platform (PHP), which is quite buggy.


> Could you set the deadlock_timeout in postgresql.conf?
It is set. Alas, it doesn't help, because there is no 
deadlock condition. (Deadlock means two transactions cannot 
continue because each tries to set lock conflicting with 
already acquired by another, mutually.)

Thank you,
Konstantin Pelepelin


> > In situations where client application is unstable
> > there is very big chance that it will enter
> > transaction, acquire very restrictive locks and then
> > hang. This means: connection is not lost, but nothing
> > happens and any transaction in conflict with those
> > locks will hang too. statement_timeout can help to
> > detect this situation, but not to solve, when most of
> > applications fail. Currently the only way I see to
> > solve is to kill locked postgres backend, or (more
> > difficult) find hanging client among hundreds of
> > similar.
> > I work with web-application (Apache-mod_php-Postgres),
> > where PHP hangs every 100000 requests or so, which
> > means from a day to some weeks between whole system
> > hangs.
> > I could set up a cron job which will kill every
> > postgres backend "idle in transaction" which was
> > started more than 10 minutes ago, but I can have
> > accident problems with some long-running tasks. The
> > best solution I see is having an (per session)
> > idle_transaction_timeout or idle_connection_timeout
> > setting. But there is no such thing in postgres!
> > Scanning through list I've not found any workaround.
> > Looking into TODO I don't see any plans for it.
> > Is it too difficult implementing or I've missed
> > something or are there workarounds?




view thread (5+ messages)  latest in thread

Message-ID: <200407121326.41770.checat@pochtamt.ru>
Permalink:  ../200407121326.41770.checat@pochtamt.ru/
Also on:    postgresql.org/message-id/200407121326.41770.checat@pochtamt.ru

 · 

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 from your mail client; the link fills in
  the headers below: reply via email
  Some clients, webmail especially, drop In-Reply-To and References from
  a link, and the reply then starts a new thread: use the mbox below.

  To: pgsql-admin@postgresql.org
  Cc: checat@pochtamt.ru, mike@thegodshalls.com
  Subject: Re: are there ways for 'idle timeout'?
  In-Reply-To: <200407121326.41770.checat@pochtamt.ru>
  References: <opsavj6izuhocrah@cathost.of.dtf.ru> <1089430151.10763.17.camel@localhost.localdomain> <200407121326.41770.checat@pochtamt.ru>

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

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