pg.ddx.io  pgsql-admin@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: Uwe C. Schroeder <uwe@oss4u.com>
To: pgsql-admin@postgresql.org
Cc: Jeff Kirby <Jeff.Kirby@wicourts.gov>
Subject: Re: Vacuum Full Analyze Stalled
Date: Mon, 3 Oct 2005 09:47:14 -0700
Message-ID: <200510030947.14676.uwe@oss4u.com> (raw)
In-Reply-To: <s33fc7c0.010@gwmta.wicourts.gov>
References: <s33fc7c0.010@gwmta.wicourts.gov>


On the linux box check if any application left a "idle in transaction" backend 
around.
just 
ps -ef | grep transaction

if that shows any postgres process then you most likely found the culprit.
An application that's still running can have a bug that starts a transaction 
which never finishes for some reason (commit or rollback it never called).
I had a couple of those errors in my app - always leading to a stuck vacuum 
full. If that's the case just kill the backend in question and the vauum will 
continue. Don't kill -9 it!!
After that try to figure out what caused the application not to commit or 
rollback a transaction. It's usually an error handler of some sort.

just an idea.

UC

On Sunday 02 October 2005 09:42, Jeff Kirby wrote:
> I'm going to attempt to do my best in describing the situation... so please
> forgive any ignorance on my part since I am very new to the Postgres
> community.
>
> We have 2 identical Postgres databases, one on a Windows 2003 platform, and
> the other on a SUSe 9.3 Linux platform (both platforms have identical
> hardware).  They both have the same schema, the same configuration settings
> (as far as I can tell), and relatively the same amount of data (approx
> 32GB).  OK... here is the scoop, I started a "vacuum full analyze verbose"
> on both last night... the windows box completed in about 1 1/2 hours... the
> Linux box however is still chugging away this morning... and appears to be
> stuck on vacuuming "pg_constraint_contypid_index".  How do I know... well I
> don't really... I'm inferring based on the order of the log output on the
> Windows box.
>
> So the question is... has anyone experienced this problem in the past?  If
> so, what do you recommend to rectify the problem?  Is it normal to be
> running for more than 12 hours on a table that only has 35 rows?  Do you
> need me to attach the log output thus far from both the Windows and Linux
> box?  Any help here would be greatly appreciated.  Any commands you can
> suggest to run on the Linux machine... no problem.
>
> Jeffrey Kirby
> CCAP Web Team
> jeff.kirby@wicourts.gov
> 608-264-6253
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 3: Have you checked our extensive FAQ?
>
>                http://www.postgresql.org/docs/faq


--
Open Source Solutions 4U, LLC	2570 Fleetwood Drive
Phone:  +1 650 872 2425		San Bruno, CA 94066
Cell:   +1 650 302 2405		United States
Fax:    +1 650 872 2417



view thread (17+ messages)  latest in thread

Message-ID: <200510030947.14676.uwe@oss4u.com>
Permalink:  ../200510030947.14676.uwe@oss4u.com/
Also on:    postgresql.org/message-id/200510030947.14676.uwe@oss4u.com

 · 

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: uwe@oss4u.com, Jeff.Kirby@wicourts.gov
  Subject: Re: Vacuum Full Analyze Stalled
  In-Reply-To: <200510030947.14676.uwe@oss4u.com>
  References: <s33fc7c0.010@gwmta.wicourts.gov> <200510030947.14676.uwe@oss4u.com>

* 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