Thread: What could cause total crash of a postgres installation?

What could cause total crash of a postgres installation?

From
"Peter Alberer"
Date:
Hi,

Since about three days I am having strange problems with my postgresl
installation. Today in the morning at about 2am the database crashed
completely. I had the postmaster running at debug level 2 but no hint
can be found in the log. The server seemed to crash without notice.

2002-11-29 02:39:01 [3294]   DEBUG:  StartTransactionCommand
2002-11-29 02:39:01 [3294]   DEBUG:  query: select preface as
preface_txt,abstract as abstract_txt,display_type,section_auto_
numbering,lr_book_revision_id as revision_id from lr_books where
lr_book_revision_id = '115876';
2002-11-29 02:39:01 [3294]   DEBUG:  ProcessQuery
2002-11-29 02:39:01 [3294]   DEBUG:  CommitTransactionCommand
2002-11-29 02:39:01 [3294]   DEBUG:  QUERY STATISTICS
! system usage stats:
!       0.001286 elapsed 0.000000 user 0.000000 system sec
!       [456.870000 user 12.170000 sys total]
!       0/0 [0/0] filesystem blocks in/out
!       0/0 [57092/25680] page faults/reclaims, 0 [0] swaps
!       0 [0] signals rcvd, 0/0 [0/0] messages rcvd/sent
!       0/0 [0/0] voluntary/involuntary context switches
! postgres usage stats:
!       Shared blocks:          0 read,          0 written, <- last
entry

At that time more or less nothing was going on in the system. I was
already collection all kinds of log information, but I can find nothing
unusual. I could start the database with no problems.

What kind of problems could cause such a crash? Could bad memory
settings (sort, buffers) cause something like that? What kind of server
statistics should I look at?

Many TIA,

peter


Re: What could cause total crash of a postgres installation?

From
"Shridhar Daithankar"
Date:
On 29 Nov 2002 at 11:06, Peter Alberer wrote:

> At that time more or less nothing was going on in the system. I was
> already collection all kinds of log information, but I can find nothing
> unusual. I could start the database with no problems.

Faulty hardware, specially disk and RAM are usual culprits.. If postgresql
crashes due to software bug(which should be rare) it should leave enough hints
in logs..

 Just check out..

Bye
 Shridhar

--
Lie, n.:    A very poor substitute for the truth, but the only one    discovered to
date.


Re: What could cause total crash of a postgres installation?

From
Tom Lane
Date:
"Peter Alberer" <h9351252@obelix.wu-wien.ac.at> writes:
> Since about three days I am having strange problems with my postgresl
> installation. Today in the morning at about 2am the database crashed
> completely.

You haven't actually shown us any evidence of a crash ... also, if the
postmaster is still alive then I'd not call it a "complete crash".
Please provide more details.

            regards, tom lane

Re: What could cause total crash of a postgres installation?

From
Medi Montaseri
Date:
I'hve also had some crashes that circle around memory management
particularly free-ing
memory.....

In my case it has to to with a functiion called pfree() which is a macro
on top of free().

Peter Alberer wrote:

>Hi,
>
>Since about three days I am having strange problems with my postgresl
>installation. Today in the morning at about 2am the database crashed
>completely. I had the postmaster running at debug level 2 but no hint
>can be found in the log. The server seemed to crash without notice.
>
>2002-11-29 02:39:01 [3294]   DEBUG:  StartTransactionCommand
>2002-11-29 02:39:01 [3294]   DEBUG:  query: select preface as
>preface_txt,abstract as abstract_txt,display_type,section_auto_
>numbering,lr_book_revision_id as revision_id from lr_books where
>lr_book_revision_id = '115876';
>2002-11-29 02:39:01 [3294]   DEBUG:  ProcessQuery
>2002-11-29 02:39:01 [3294]   DEBUG:  CommitTransactionCommand
>2002-11-29 02:39:01 [3294]   DEBUG:  QUERY STATISTICS
>! system usage stats:
>!       0.001286 elapsed 0.000000 user 0.000000 system sec
>!       [456.870000 user 12.170000 sys total]
>!       0/0 [0/0] filesystem blocks in/out
>!       0/0 [57092/25680] page faults/reclaims, 0 [0] swaps
>!       0 [0] signals rcvd, 0/0 [0/0] messages rcvd/sent
>!       0/0 [0/0] voluntary/involuntary context switches
>! postgres usage stats:
>!       Shared blocks:          0 read,          0 written, <- last
>entry
>
>At that time more or less nothing was going on in the system. I was
>already collection all kinds of log information, but I can find nothing
>unusual. I could start the database with no problems.
>
>What kind of problems could cause such a crash? Could bad memory
>settings (sort, buffers) cause something like that? What kind of server
>statistics should I look at?
>
>Many TIA,
>
>peter
>
>
>---------------------------(end of broadcast)---------------------------
>TIP 5: Have you checked our extensive FAQ?
>
>http://www.postgresql.org/users-lounge/docs/faq.html
>
>




Re: What could cause total crash of a postgres

From
Neil Conway
Date:
On Mon, 2002-12-02 at 15:05, Medi Montaseri wrote:
> I'hve also had some crashes that circle around memory management
> particularly free-ing memory.....

AFAICS these two crashes have nothing to do with one another.

> In my case it has to to with a functiion called pfree() which is a macro
> on top of free().

(BTW, pfree() is *not* a macro on top of free().)

You've posted the stack backtrace from your 7.2.3 crash in the past, but
AFAIK no one has followed up on it. Could you see if you can reproduce
the problem in 7.3, and if you can get a backtrace from a core dump
produced by a version of PostgreSQL with debugging symbols? (./configure
--enable-debug).

Cheers,

Neil
--
Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC




Re: What could cause total crash of a postgres installation?

From
"Peter Alberer"
Date:
Well, as far as my crash-problem is concerned, I have looked thru
various log files in /var/log, examined sar data and could not find a
single hint on what could be wrong. I have put postmaster under
supervise (daemontools) and removed debug logging again. Since Friday
everything is running without problems.

As Tom Lane told me there is no chance that any problem with a single
query could take out the postmaster and all of the backends at once. So
maybe that really was a system-level problem...

Thanks to all of you who responded to my request!

peter


>-----Ursprüngliche Nachricht-----
>Von: pgsql-general-owner@postgresql.org [mailto:pgsql-general-
>owner@postgresql.org] Im Auftrag von Medi Montaseri
>Gesendet: Montag, 02. Dezember 2002 21:06
>An: Peter Alberer
>Cc: pgsql-general@postgresql.org
>Betreff: Re: [GENERAL] What could cause total crash of a postgres
>installation?
>
>I'hve also had some crashes that circle around memory management
>particularly free-ing
>memory.....
>
>In my case it has to to with a functiion called pfree() which is a
macro
>on top of free().
>
>Peter Alberer wrote:
>
>>Hi,
>>
>>Since about three days I am having strange problems with my postgresl
>>installation. Today in the morning at about 2am the database crashed
>>completely. I had the postmaster running at debug level 2 but no hint
>>can be found in the log. The server seemed to crash without notice.
>>
>>2002-11-29 02:39:01 [3294]   DEBUG:  StartTransactionCommand
>>2002-11-29 02:39:01 [3294]   DEBUG:  query: select preface as
>>preface_txt,abstract as abstract_txt,display_type,section_auto_
>>numbering,lr_book_revision_id as revision_id from lr_books where
>>lr_book_revision_id = '115876';
>>2002-11-29 02:39:01 [3294]   DEBUG:  ProcessQuery
>>2002-11-29 02:39:01 [3294]   DEBUG:  CommitTransactionCommand
>>2002-11-29 02:39:01 [3294]   DEBUG:  QUERY STATISTICS
>>! system usage stats:
>>!       0.001286 elapsed 0.000000 user 0.000000 system sec
>>!       [456.870000 user 12.170000 sys total]
>>!       0/0 [0/0] filesystem blocks in/out
>>!       0/0 [57092/25680] page faults/reclaims, 0 [0] swaps
>>!       0 [0] signals rcvd, 0/0 [0/0] messages rcvd/sent
>>!       0/0 [0/0] voluntary/involuntary context switches
>>! postgres usage stats:
>>!       Shared blocks:          0 read,          0 written, <- last
>>entry
>>
>>At that time more or less nothing was going on in the system. I was
>>already collection all kinds of log information, but I can find
nothing
>>unusual. I could start the database with no problems.
>>
>>What kind of problems could cause such a crash? Could bad memory
>>settings (sort, buffers) cause something like that? What kind of
server
>>statistics should I look at?
>>
>>Many TIA,
>>
>>peter
>>
>>
>>---------------------------(end of
broadcast)---------------------------
>>TIP 5: Have you checked our extensive FAQ?
>>
>>http://www.postgresql.org/users-lounge/docs/faq.html
>>
>>
>
>
>
>
>---------------------------(end of
broadcast)---------------------------
>TIP 5: Have you checked our extensive FAQ?
>
>http://www.postgresql.org/users-lounge/docs/faq.html