Re: could not rename xlog (was: BUG #2712)

another lockup, this time due to pgstat_write_statsfile :-(

2006-10-24 17:01:17 [5412] LOG:  XX000: could not rename temporary
statistics file "global/pgstat.tmp" to "global/pgstat.stat": A blocking
operation was interrupted by a call to WSACancelBlockingCall.
2006-10-24 17:01:17 [5412] LOCATION:  pgstat_write_statsfile, pgstat.c:2008
2006-10-24 17:23:23 [3280] LOG:  00000: received fast shutdown request
2006-10-24 17:23:23 [3280] LOCATION:  pmdie, postmaster.c:1903
2006-10-24 17:23:23 [3280] LOG:  00000: aborting any active transactions
2006-10-24 17:23:23 [3280] LOCATION:  pmdie, postmaster.c:1910
2006-10-24 17:23:23 [3468] FATAL:  57P01: terminating connection due to
administrator command
2006-10-24 17:23:23 [3468] LOCATION:  ProcessInterrupts, postgres.c:2465

this is what filemon reports (~10 times a second):

17:23:18 postgres.exe:1432 OPEN
D:\DB\PostgreSQL-8.2\data\pg_xlog\0000000100000004000000DB DELETE PEND
Options: Open  Access: 00110080
17:23:18 postgres.exe:1432 OPEN
D:\DB\PostgreSQL-8.2\data\pg_xlog\0000000100000004000000DB DELETE PEND
Options: Open  Access: 00110080
17:23:18 postgres.exe:1432 OPEN
D:\DB\PostgreSQL-8.2\data\pg_xlog\0000000100000004000000DB DELETE PEND
Options: Open  Access: 00110080

sorry for flooding. just tell me if i shall rather stop.

- thomas






----- Original Message -----
From: "Thomas H." <me@alternize.com>
To: <pgsql-bugs@postgresql.org>
Sent: Tuesday, October 24, 2006 3:15 PM
Subject: Re: [BUGS] could not rename xlog (was: BUG #2712)


>> "Peter Brant" <Peter.Brant@wicourts.gov> writes:
>>> The same problem exists in 8.1 too.  See this thread
>>> http://archives.postgresql.org/pgsql-bugs/2006-04/msg00177.php
>>> Tom and Magnus tracked down a cause, but I don't think a fix was ever
>>> implemented.
>>
>> Thomas seems to have two different issues there: the "could not rename
>> file" problem on the pg_xlog file is probably explained by the mechanism
>> we identified back then (and I'm not sure why no fix has been
>> installed)
>
> just had another total lockdown. the writer-process was trying to rename
> the C1 to CA which failed:
>
> 2006-10-24 14:27:58 [5196] LOG:  00000: could not rename file
> "pg_xlog/0000000100000004000000C1" to "pg_xlog/0000000100000004000000CA",
> continuing to try
> 2006-10-24 14:27:58 [5196] LOCATION:  pgrename, dirmod.c:142
>
> when checking the process with process explorer, it reveals that it has
> this file handles to pg_xlog open:
>
> D:\DB\PostgreSQL-8.2\data\pg_xlog
> D:\DB\PostgreSQL-8.2\data\pg_xlog\0000000100000004000000C3
>
> under normal operation, "writer process" does not seem to have a file
> handle to the xlog directory (D:\DB\PostgreSQL-8.2\data\pg_xlog)
>
> the last error log entry prior to the lockdown was about 15min, so
> probably the two problems are unrelated.
>
> unfortunately, before i could try to get more informations, i had to
> restart the pg_ctl due to the system being in half-productive mode. that
> leads me to the question: what are the chances of me being helpful with
> getting more informations? i can have the db run like that for some more
> days, but for long term i can't babysit our application and might have to
> go back to 8.1.
>
> thanks,
> thomas
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: if posting/reading through Usenet, please send an appropriate
>       subscribe-nomail command to majordomo@postgresql.org so that your
>       message can get through to the mailing list cleanly
>