Postgres Pro
Downloads
Home   >   mailing lists

Thread: DB crash after disk full

DB crash after disk full

From
Vilen Tambovtsev
Date:
Hello,
My disk went out of space, so pgsql crashed, and had some FS troubles occurred.
I fixed FS, then tried to start and got this:

  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  database system was interrupted at
2005-12-28 19:02:09 NOVT
  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  checkpoint record is at 5/FA0000E8
  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  redo record is at 5/FA0000E8; undo record is
at 0/0; shutdown TRUE
  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  next transaction ID: 74678708; next OID: 2982120
  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  next MultiXactId: 1; next MultiXactOffset: 0
  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  database system was not properly shut down;
automatic recovery in progress
  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  record with zero length at 5/FA00012C
  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  redo is not required
  - 2005-12-28 19:02:49 NOVT -  - 2747: PANIC:  could not access status of transaction
74678708
  - 2005-12-28 19:02:49 NOVT -  - 2747: DETAIL:  could not read from file "pg_clog/0047"
at offset 57344: Unknown error:
0
  - 2005-12-28 19:02:49 NOVT -  - 2745: LOG:  startup process (PID 2747) was terminated by
signal 6
  - 2005-12-28 19:02:49 NOVT -  - 2745: LOG:  aborting startup due to startup process failure
  - 2005-12-28 19:02:49 NOVT -  - 2746: LOG:  logger shutting down

Tried to use pg_resetxlog - but it has not effect :(
Is there any tool to recover or smth ?
Maybe to reset transaction log...

--
Vilen Tambovtsev

Re: DB crash after disk full

From
Vilen Tambovtsev
Date:
Got to make it work, but now, when i try to make a dump it says
pg_dump: SQL command failed
pg_dump: Error message from server: ERROR:  missing chunk number 0 for toast value 2969644
pg_dump: The command was: FETCH 100 FROM _pg_dump_cursor

tried reindex:
reindexdb: reindexing of database "hostel" failed: ERROR:  missing chunk number 0 for
toast value 2969644
and vacuum -f
vacuumdb: vacuuming of database "hostel" failed: ERROR:  missing chunk number 0 for toast
value 2970344


Vilen Tambovtsev wrote:
> Hello,
> My disk went out of space, so pgsql crashed, and had some FS troubles
> occurred.
> I fixed FS, then tried to start and got this:
>
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  database system was
> interrupted at 2005-12-28 19:02:09 NOVT
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  checkpoint record is at
> 5/FA0000E8
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  redo record is at
> 5/FA0000E8; undo record is at 0/0; shutdown TRUE
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  next transaction ID:
> 74678708; next OID: 2982120
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  next MultiXactId: 1; next
> MultiXactOffset: 0
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  database system was not
> properly shut down; automatic recovery in progress
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  record with zero length at
> 5/FA00012C
>  - 2005-12-28 19:02:49 NOVT -  - 2747: LOG:  redo is not required
>  - 2005-12-28 19:02:49 NOVT -  - 2747: PANIC:  could not access status
> of transaction 74678708
>  - 2005-12-28 19:02:49 NOVT -  - 2747: DETAIL:  could not read from file
> "pg_clog/0047" at offset 57344: Unknown error:
> 0
>  - 2005-12-28 19:02:49 NOVT -  - 2745: LOG:  startup process (PID 2747)
> was terminated by signal 6
>  - 2005-12-28 19:02:49 NOVT -  - 2745: LOG:  aborting startup due to
> startup process failure
>  - 2005-12-28 19:02:49 NOVT -  - 2746: LOG:  logger shutting down
>
> Tried to use pg_resetxlog - but it has not effect :(
> Is there any tool to recover or smth ?
> Maybe to reset transaction log...
>

--
Vilen Tambovtsev
Plesk QA Engineer
SWsoft, Inc.
E-mail: vtambovtsev@swsoft.com
ICQ UIN: 4608679

Re: DB crash after disk full

From
Tom Lane
Date:
Vilen Tambovtsev <vtambovtsev@swsoft.com> writes:
>   - 2005-12-28 19:02:49 NOVT -  - 2747: PANIC:  could not access status of transaction 74678708

What Postgres version is this?

            regards, tom lane

Re: DB crash after disk full

From
Vilen Tambovtsev
Date:
Version 8.1.1
Got to make it work, after deletin some records from table

Tom Lane wrote:
> Vilen Tambovtsev <vtambovtsev@swsoft.com> writes:
>>   - 2005-12-28 19:02:49 NOVT -  - 2747: PANIC:  could not access status of transaction 74678708
>
> What Postgres version is this?
>
>             regards, tom lane
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: explain analyze is your friend
>

--
Vilen Tambovtsev
Plesk QA Engineer
SWsoft, Inc.
E-mail: vtambovtsev@swsoft.com
ICQ UIN: 4608679