Re: Allow WAL information to recover corrupted pg_controldata

On Wed, Jun 20, 2012 at 12:40 PM, Amit Kapila <amit(dot)kapila(at)huawei(dot)com> wrote:
>>> I believe if WAL files are proper as mentioned in Alvaro's mail, the
> purposed logic should generate
>>> correct values.
>>> Do you see any problem in logic purposed in my original mail.
>>> Can I resume my work on this feature?
>
>> Maybe I'm missing your point, but... why don't you just use PITR to
>> recover from the corruption of pg_control?
>
> AFAIK PITR can be used in a scenario where there is a base back-up and we
> have archived
> the WAL files after that, now it can use WAL files to apply on base-backup.

Yes. If you want to recover the database from the media crash like the
corruption of pg_control file, you basically should take a base backup
and set up continuous archiving.

> In this scenario we don't know a point from where to start the next replay.
> So I believe it will be difficult to use PITR in this scenario.

You can find out the point from the complete pg_control file which was
restored from the backup.

If pg_control is corrupted, we can easily imagine that other database files
would also be corrupted. I wonder how many cases where only pg_control
file gets corrupted are. In that case, pg_resetxlog is unhelpful at all.
You need to use PITR, intead.

Regards,

--
Fujii Masao