Обсуждение: Postgres server crashing unexpectedly.

"umashankar narayanan" <umashankar@graffiti.net> wrote:
> Version : 8.3
>
>
>  Below is the log from the server.
>
>
> -----------------------------------------------------------------
> ------------------------------------------------------

The above is everything that showed up on your post.  Make sure
you're not attaching large images, file which can only be viewed on
a particular operating system, or anything else unusual.

Please read this and try again:

http://wiki.postgresql.org/wiki/Guide_to_reporting_problems

-Kevin

"umashankar narayanan" <umashankar@graffiti.net> wrote:

> Version : 8.3

Please give the full version and a bit more information about the
environment it's running in:

http://wiki.postgresql.org/wiki/Guide_to_reporting_problems

>  2012-03-07 11:57:20 EST LOG: server process (PID 5944) exited
> with exit code 128

Was there anything in the PostgreSQL log ahead of this to give a
clue what went wrong and caused this?  Is there anything in any of
system logs from around this time?

>  2012-03-07 11:57:20 EST LOG: terminating any other active server
> processes

This is normal -- when one process crashes this hard, it can leave
shared memory in a corrupted state, so PostgreSQL restarts to get a
clean environment with minimal down time.

> [each client process logging its termination]

>  2012-03-07 11:57:21 EST LOG: all server processes terminated;
> reinitializing

OK, it's trying to restart so clients can re-connect and continue.

>  2012-03-07 11:57:22 EST FATAL: pre-existing shared memory block
> is still in use
>  2012-03-07 11:57:22 EST HINT: Check if there are any old server
> processes still running, and terminate them.

Something is still hanging on to the PostgreSQL shared memory block.

Are any postgres processes lingering?  Are you running anything else
which might be grabbing the same shared memory block?

-Kevin

On Thu, Mar 8, 2012 at 12:23 PM, umashankar narayanan
<umashankar@graffiti.net> wrote:
> Kevin,
>
> The full version of Postgres is 8.3.1. It is running on Microsoft Windows
> 2003 Standard 32-bit on VMWare.
> On the system logs around this time, the only message that we receive is
> that the PID crashed.
>
> All that I see on the postgres log before the crash is the below message:
>
> 2012-03-07 11:56:12 EST LOG:  loaded library
> "$libdir/plugins/plugin_debugger.dll"
>
> No other postgres processes are lingering. I did confirm that no other batch
> jobs or monitoring jobs were running during this time, that would grab the
> same shared memory block.
>
> Thanks!
> Umashankar
>
>
>
>
>
> ----- Original Message -----
>
> From: Kevin Grittner
>
> Sent: 03/08/12 01:25 PM
>
> To: umashankar narayanan, pgsql-admin@postgresql.org
>
> Subject: Re: [ADMIN] Postgres server crashing unexpectedly.
>
>
>
> "umashankar narayanan" <umashankar@graffiti.net> wrote:
>
>> Version : 8.3
>
> Please give the full version and a bit more information about the
> environment it's running in:
>
> http://wiki.postgresql.org/wiki/Guide_to_reporting_problems
>
>>  2012-03-07 11:57:20 EST LOG: server process (PID 5944) exited
>> with exit code 128
>
> Was there anything in the PostgreSQL log ahead of this to give a
> clue what went wrong and caused this?  Is there anything in any of
> system logs from around this time?
>
>>  2012-03-07 11:57:20 EST LOG: terminating any other active server
>> processes
>
> This is normal -- when one process crashes this hard, it can leave
> shared memory in a corrupted state, so PostgreSQL restarts to get a
> clean environment with minimal down time.
>
>> [each client process logging its termination]
>
>>  2012-03-07 11:57:21 EST LOG: all server processes terminated;
>> reinitializing
>
> OK, it's trying to restart so clients can re-connect and continue.
>
>>  2012-03-07 11:57:22 EST FATAL: pre-existing shared memory block
>> is still in use
>>  2012-03-07 11:57:22 EST HINT: Check if there are any old server
>> processes still running, and terminate them.
>
> Something is still hanging on to the PostgreSQL shared memory block.
>
> Are any postgres processes lingering?  Are you running anything else
> which might be grabbing the same shared memory block?

If I remember correctly this bug was fixed in later windows versions.
Can you try upgrading to 8.3.latest?

[Please don't top-post.  http://idallen.com/topposting.html ]

"umashankar narayanan" <umashankar@graffiti.net> wrote:
> "Kevin Grittner" <Kevin.Grittner@wicourts.gov> wrote:
>> "umashankar narayanan" <umashankar@graffiti.net> wrote:

>> Please give the full version

> The full version of Postgres is 8.3.1.

You are missing four years of bug fixes for 8.3.

http://www.postgresql.org/support/versioning/

To see what fixes you are missing, look at the links to release
notes for 8.3.2 to 8.3.18:

http://www.postgresql.org/docs/8.3/static/release.html

Please either update to a recent bug-fix release of 8.3 or review
all of these pages closely to be sure that your problem couldn't
have been caused by any of the already-fixed bugs.

>>> 2012-03-07 11:57:20 EST LOG: server process (PID 5944) exited
>>> with exit code 128
>>
>> Was there anything in the PostgreSQL log ahead of this to give a
>> clue what went wrong and caused this?

> All that I see on the postgres log before the crash is the below
> message:
>
> 2012-03-07 11:56:12 EST LOG: loaded library
> "$libdir/plugins/plugin_debugger.dll"

That's about one minute before the crash.  If that's not what caused
the crash, it's a heck of a coincidence.

>>> 2012-03-07 11:57:22 EST FATAL: pre-existing shared memory block
>>> is still in use
>>> 2012-03-07 11:57:22 EST HINT: Check if there are any old server
>>> processes still running, and terminate them.
>>
>> Something is still hanging on to the PostgreSQL shared memory
>> block.
>>
>> Are any postgres processes lingering?  Are you running anything
>> else which might be grabbing the same shared memory block?

> No other postgres processes are lingering. I did confirm that no
> other batch jobs or monitoring jobs were running during this time,
> that would grab the same shared memory block.

Perhaps the debugger still had some process active and attached to
shared memory at the point that this was logged, but has since
released the memory?  *Something* was still using PostgreSQL shared
memory at 2012-03-07 11:57:22 EST.

-Kevin

Scott Marlowe <scott.marlowe@gmail.com> writes:
> On Thu, Mar 8, 2012 at 12:23 PM, umashankar narayanan
> <umashankar@graffiti.net> wrote:
>> The full version of Postgres is 8.3.1. It is running on Microsoft Windows
>> 2003 Standard 32-bit on VMWare.

> If I remember correctly this bug was fixed in later windows versions.

see release notes for 8.3.8 ...