Re: [BUGS] Loosing files after backend crash - Mailing list pgsql-hackers

From Tom Lane
Subject Re: [BUGS] Loosing files after backend crash
Date
Msg-id 12826.986423898@sss.pgh.pa.us
Whole thread Raw
List pgsql-hackers
KS <ks@tcnet.ru> writes:
> Well, bad news. After a few more tries to crash the backend, the whole
> postmaster crashed and didn't rise back.
> It fails to start up reporting "Apr  4 18:53:05 wale postgres[71618]: [9]
> FATAL 2:  XLogWrite: write request is past end of log" to syslog.

Ugh.

> And the last line of errlog sounds like "/usr/local/pgsql/bin/postmaster:
> Startup proc 72905 exited with status 512 - abort"
> I wanted to ask, if I need to re-initdb or there are some other ways to fix
> the problem?

You can use contrib/pg_resetxlog to remove the damaged WAL log, which
will allow the postmaster to start up.  If you have the space, please
first save the contents of $PGDATA (the whole tree if possible, else at
least pg_xlog directory) for later analysis.  Note that the database
might not be completely consistent after you zap the WAL log --- safest
bet would be to dump the data, look it over for problems, then initdb
and restore.

> Shall I send you it's code? (It's 23 Kbytes big).

Please.  23K is no problem.

BTW, exactly which version of Postgres are you working with --- is it a
CVS snapshot, or a beta or RC release, and if so which one?
        regards, tom lane


pgsql-hackers by date:

Previous
From: Martín Marqués
Date:
Subject: Re: Configure problems on Solaris 2.7, pgsql 7.02 and 7.03
Next
From: Thomas Lockhart
Date:
Subject: Re: All's quiet ... RC3 packaging ...