Thread: out of memory ...

out of memory ...

From
"Jan-Peter Seifert"
Date:
Hello,

how fatal is it for the server/data if a server runs out of memory?

Happened on a Windows server some time ago. A transaction got too large. In the log it began with:

------------
TopMemoryContext: 37171728 total in 4046 blocks; 67640 free (4060 chunks); 37104088 used
  TopTransactionContext: 24576 total in 2 blocks; 18848 free (17 chunks); 5728 used
    CurTransactionContext: 8192 total in 1 blocks; 8176 free (1 chunks); 16 used
...
------------

Although all other clients had been closed no new connection was possible to terminate the session responsible for the
outof memory problem. I guess it's because this transaction had been started by a superuser ... 

------------

In the end the server was shut down via the service manager.

On the next start of the server the log contained the following lines:

------------
LOG:  database system was interrupted; last known up at 2010-03-26 11:22:36 CET
LOG:  database system was not properly shut down; automatic recovery in progress
LOG:  record with zero length at 1F/D2454CD0
------------

The data still seemed to be okay - all the other transactions made before seem to be in the database.

Is it okay to continue using the database(s) or is there a risk of hidden data corruption (fsync is on)?

Could you tell me, please?

Thank you very much,

Peter


--
GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT!
Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01

Re: out of memory ...

From
Cédric Villemain
Date:
2010/5/3 Jan-Peter Seifert <Jan-Peter.Seifert@gmx.de>:
> Hello,
>
> how fatal is it for the server/data if a server runs out of memory?
>
> Happened on a Windows server some time ago. A transaction got too large. In the log it began with:
>
> ------------
> TopMemoryContext: 37171728 total in 4046 blocks; 67640 free (4060 chunks); 37104088 used
>  TopTransactionContext: 24576 total in 2 blocks; 18848 free (17 chunks); 5728 used
>    CurTransactionContext: 8192 total in 1 blocks; 8176 free (1 chunks); 16 used
> ...
> ------------
>
> Although all other clients had been closed no new connection was possible to terminate the session responsible for
theout of memory problem. I guess it's because this transaction had been started by a superuser ... 
>
> ------------
>
> In the end the server was shut down via the service manager.
>
> On the next start of the server the log contained the following lines:
>
> ------------
> LOG:  database system was interrupted; last known up at 2010-03-26 11:22:36 CET
> LOG:  database system was not properly shut down; automatic recovery in progress
> LOG:  record with zero length at 1F/D2454CD0
> ------------
>
> The data still seemed to be okay - all the other transactions made before seem to be in the database.
>
> Is it okay to continue using the database(s) or is there a risk of hidden data corruption (fsync is on)?

okay to continue if your filesystem didn't have issues (if postgres
start, then it is ok from its side)

>
> Could you tell me, please?
>
> Thank you very much,
>
> Peter
>
>
> --
> GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT!
> Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01
>
> --
> Sent via pgsql-admin mailing list (pgsql-admin@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-admin
>



--
Cédric Villemain

Re: out of memory ...

From
Jan-Peter Seifert
Date:
Hello,

>> ------------
>> LOG:  database system was interrupted; last known up at 2010-03-26 11:22:36 CET
>> LOG:  database system was not properly shut down; automatic recovery in progress
>> LOG:  record with zero length at 1F/D2454CD0
>> ------------
>>
>> The data still seemed to be okay - all the other transactions made before seem to be in the database.
>>
>> Is it okay to continue using the database(s) or is there a risk of hidden data corruption (fsync is on)?
>
> okay to continue if your filesystem didn't have issues (if postgres
> start, then it is ok from its side)

Sounds good.

Thank you very much for your reply.

Peter