Re: Recover from corrupted database due to failing disk

On 11/04/2016 03:20 AM, Gionatan Danti wrote:
>
>
> On 03/11/2016 14:20, Adrian Klaver wrote:
>>
>> The above does not make sense. You are having to recover because there
>> was no backup and now you want to go forward without doing a backup?
>>
>
> Hi Adrian, no, I don't want go forward without backups ;)
> Actually, the *first* thing I did after the vacuum completed was a full
> cluster backup (via pg_dumpall), and I scheduled nightly backups as well.
>
> Problem is this customer does not have another server were backups can
> be restored and the entire production database migrated. In short, the
> two possibilities I have are:
>
> 1) execute the vacuum (done), schedule regular dumps (done) and, if
> something goes wrong, recover from backups;
>
> 2) execute the vacuum (done), do a manual backup (done), reinit
> (remove/recreate) the entire cluster (not done) and restore from backups
> (not done).
>
> I strongly prefer to execute n.2 on another machine, so that production
> is not impacted while the recovered backup can be througly tested.
> If/when the backups are validated, I want to migrate all clients to the
> new server (with RAID1 in place), and dismiss the old one.
>
> Unfortuntaly I am working with incredible constrains from customer side;
> even buying two SAS disks seems a problem. Moreover, as an external
> consultant, I have basically no decision/buying power :|
> What I can do (and I did) is to raise a very big red flag and let others
> decide what to do.

Ouch, understood. Good luck!

>
> The good thing is that zero_damaged_pages and vacuum did their works, as
> now the database can be dumped and vacuumed with no (apparent) problems.
>
> Thanks.
>


--
Adrian Klaver
adrian.klaver@aklaver.com