Lists: | pgsql-bugs |
---|
From: | Robert Schnabel <schnabelr(at)missouri(dot)edu> |
---|---|
To: | pgsql-bugs(at)postgresql(dot)org |
Subject: | pgstat wait timeout |
Date: | 2009-07-08 15:13:45 |
Message-ID: | 4A54B7A9.8070601@missouri.edu |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Lists: | pgsql-bugs |
I've been having an issue for about 5 months now and after upgrading to
8.4 did not fix it I figured I'd try this list. I tried the perform
list a while back but no bites.
My system: Windows XP64 SP2, dual Opteron 2218, 12G ram, PostgreSQL
8.4. I am the only person using the postgres server and I only have one
db. Yesterday I dumped from 8.3.7, installed 8.4 and restored my db to
8.4. Everything works fine except the problem that I've had since about
January. I didn't have this problem until I upgraded to 8.3.6 (I
think). It did the same thing with 8.3.7. Here is what's happening...
From a fresh server restart I have one process that will run one
processor 100% essentially forever. (I've let it go for about a week
just to see if it would ever stop and it didn't). Attached is a text
dump from ProcessExplorer. I have highlighted the relevant lines with
****. Also attached is the current postgresql log after a server
restart. You can see from the log that about every 70 seconds there is
an entry "WARNING: pgstat wait timeout". This will go on forever.
From the ProcessExplorer file, the offending process is 2384. From
within ProcessExplorer, if I suspend 2384 the log entries do not cease.
>From within ProcessExplorer, if I attach the debugger it will
immediately terminate the process and then restart with a new process
id. For the last 5 months since I noticed the problem I have resorted
to using ProcessExplorer to suspend the process, thus freeing up a CPU,
and just let the log files get cluttered. I assume this is a bug?? Any
help would be greatly appreciated.
Bob
Attachment | Content-Type | Size |
---|---|---|
postgresql-2009-07-08_093646.log | text/plain | 961 bytes |
PE_postgres.txt | text/plain | 9.6 KB |
From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Robert Schnabel <schnabelr(at)missouri(dot)edu> |
Cc: | pgsql-bugs(at)postgresql(dot)org |
Subject: | Re: pgstat wait timeout |
Date: | 2009-07-08 15:43:22 |
Message-ID: | 22055.1247067802@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Lists: | pgsql-bugs |
Robert Schnabel <schnabelr(at)missouri(dot)edu> writes:
> From a fresh server restart I have one process that will run one
> processor 100% essentially forever. (I've let it go for about a week
> just to see if it would ever stop and it didn't). Attached is a text
> dump from ProcessExplorer. I have highlighted the relevant lines with
> ****. Also attached is the current postgresql log after a server
> restart. You can see from the log that about every 70 seconds there is
> an entry "WARNING: pgstat wait timeout". This will go on forever.
Apparently the stats collector has gotten stuck in a loop and is failing
to update the stats file. The 'wait timeout' bleats are probably coming
from the autovacuum launcher process, which is waiting for an up-to-date
stats file to appear. Can you get a stack trace from the stats
collector process?
regards, tom lane
From: | Robert Schnabel <schnabelr(at)missouri(dot)edu> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | pgsql-bugs(at)postgresql(dot)org |
Subject: | Re: pgstat wait timeout |
Date: | 2009-07-24 13:58:32 |
Message-ID: | 4A69BE08.9040707@missouri.edu |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Lists: | pgsql-bugs |
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
<br>
Tom Lane wrote:
<blockquote cite="mid:22055(dot)1247067802(at)sss(dot)pgh(dot)pa(dot)us" type="cite">
<pre wrap="">Robert Schnabel <a class="moz-txt-link-rfc2396E" href="mailto:schnabelr(at)missouri(dot)edu"><schnabelr(at)missouri(dot)edu></a> writes:
</pre>
<blockquote type="cite">
<pre wrap=""> From a fresh server restart I have one process that will run one
processor 100% essentially forever. (I've let it go for about a week
just to see if it would ever stop and it didn't). Attached is a text
dump from ProcessExplorer. I have highlighted the relevant lines with
****. Also attached is the current postgresql log after a server
restart. You can see from the log that about every 70 seconds there is
an entry "WARNING: pgstat wait timeout". This will go on forever.
</pre>
</blockquote>
<pre wrap=""><!---->
Apparently the stats collector has gotten stuck in a loop and is failing
to update the stats file. The 'wait timeout' bleats are probably coming
from the autovacuum launcher process, which is waiting for an up-to-date
stats file to appear. Can you get a stack trace from the stats
collector process?
regards, tom lane
</pre>
</blockquote>
Sorry for the delayed response to this but I just had too many fires to
put out. In the mean time, this problem seems to have cured itself. I
rebooted my computer yesterday (Windows XP), the postgres service
started and I'm no longer having an issue. I had not rebooted since I
upgraded to 8.4 but it seems to have done the trick. Previously in
8.3.7 I would have this problem every time I would reboot.<br>
<br>
Bob<br>
<br>
</body>
</html>
Attachment | Content-Type | Size |
---|---|---|
unknown_filename | text/html | 1.8 KB |