Re: virtualized snapshots and PITR - Mailing list pgsql-admin

From Magnus Hagander
Subject Re: virtualized snapshots and PITR
Date
Msg-id CABUevEz262UDf+4_CL7VVPBGiVsj1bmiSARFPF5Mxr3hGjdgkQ@mail.gmail.com
Whole thread Raw
In response to Re: virtualized snapshots and PITR  (PropAAS DBA <dba@propaas.com>)
List pgsql-admin
On Thu, Dec 14, 2017 at 4:29 PM, PropAAS DBA <dba@propaas.com> wrote:


On 12/14/2017 01:19 AM, Laurenz Albe wrote:
PropAAS DBA wrote:
we have a client using veeam to do vmware snapshots of the db servers,
unfortunately as far as I can tell any restore of a veeam snapshot also
restores the state, which means the db gets restored in an online state.
As far as I know, once the db is online we cannot force it back into
recovery mode to do a PITR recovery, correct?

Anyone know of a way to do PITR recoveries with veeam?
It is fairly simple.

Before the snapshot, run

    SELECT pg_start_backup('label of your choice', TRUE);

After the snapshot, run

    SELECT pg_stop_backup();

Yours,
Laurenz Albe

However restoring the snapshot with veeam will restore the db to a running/online state, which means we cannot force it to go back into recovery mode yes?

If your snapshots include the memory/cpu state, then you cannot use them for point in time recovery. It only works if what you snapshot is the disk.

If the snapshots include memory/cpu state they should be usable without the WAL. *should* being the operative word. But not for PITR.

I don't know veeam specifically, though, so I don't know how it works there. Unless they specifically document support for it, I wouldn't use it for PITR under any conditions.


--

pgsql-admin by date:

Previous
From: PropAAS DBA
Date:
Subject: Re: virtualized snapshots and PITR
Next
From: Guillaume Lelarge
Date:
Subject: Re: No stats after promoting standby?