Re: Trying to handle db corruption 9.6

On Mon, May 20, 2019 at 08:20:33PM +0300, Mariel Cherkassky wrote:
> Hey Greg,
> Basically my backup was made after the first pg_resetxlog so I was wrong.

Bummer.

> However, the customer had a secondary machine that wasn't synced for a
> month. I have all the walls since the moment the secondary went out of
> sync. Once I started it I hoped that it will start recover the wals and
> fill the gap. However I got an error in the secondary :         
>  2019-05-20 10:11:28 PDT  19021  LOG:  entering standby mode
> 2019-05-20 10:11:28 PDT  19021  LOG:  invalid primary checkpoint record
> 2019-05-20 10:11:28 PDT  19021  LOG:  invalid secondary checkpoint link in
> control file
> 2019-05-20 10:11:28 PDT  19021  PANIC:  could not locate a valid
> checkpoint record
> 2019-05-20 10:11:28 PDT  19018  LOG:  startup process (PID 19021) was
> terminated by signal 6: Aborted
> 2019-05-20 10:11:28 PDT  19018  LOG:  aborting startup due to startup
> process failure
> 2019-05-20 10:11:28 PDT  19018  LOG:  database system is shut down.       
>                     I checked my secondary archive dir and pg_xlog dir and
> it seems that the restore command doesnt work. My restore_command:      
> restore_command = 'rsync -avzhe ssh
> postgres(at)x(dot)x(dot)x(dot)x:/var/lib/pgsql/archive/%f /var/lib/pgsql/archive/%f ;
> gunzip < /var/lib/pgsql/archive/%f > %p'
> archive_cleanup_command = '/usr/pgsql-9.6/bin/pg_archivecleanup
> /var/lib/pgsql/archive %r'

Well, when you say it does not work, why do you think so? Does it print
some error, or what? Does it even get executed? It does not seem to be
the case, judging by the log (there's no archive_command message).

How was the "secondary machine" created? You said you have all the WAL
since then - how do you know that?

regards

--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services