From: | chiru r <chirupg(at)gmail(dot)com> |
---|---|
To: | Stephen Frost <sfrost(at)snowman(dot)net> |
Cc: | Forums postgresql <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: PgBackRest question? |
Date: | 2017-12-15 21:36:16 |
Message-ID: | CA+RSxMgR-avEre8eCAkc795iNUiq_S3cmrvS1Lz_Lw6vZxdhPg@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Hi All,
Thanks,I am thinking about a specific recovery case.
Lets assume Heavy transactional system we configured.
It is generating WAL 2000/hr and recycling automatically in pg_wal
directory.
QA :
Sunday -- 11 PM -- Full backup done.
Monday -- 11 PM -- Differential Backup done
Tuesday-- 10 AM incremental backup is done
Note : Every 2 hrs incremental backup scheduled on system.
For example, if we want to restore to DEV server:
We want to recover database as of 11:30 AM Tuesday on DEV server using QA
backups.
Is it possible to restore using pgbackrest tool for this scenario?.
How pgbackrest keeps track of transactions since the last backup? Where it
stores transaction information for recovery ?.
Thanks,
Chiru
On Wed, Dec 13, 2017 at 5:45 PM, Stephen Frost <sfrost(at)snowman(dot)net> wrote:
> Greetings,
>
> * chiru r (chirupg(at)gmail(dot)com) wrote:
> > The pgbackrest.conf configuration file has the option *repo-path* to
> mention
> > the mount point to store the backups and wal archive files.
>
> That's correct, that's where the backups and the WAL are stored.
>
> > We are looking for an option, Can we store online backups and wal archive
> > files separately in different directories?.
>
> That's an interesting requirement- why would you need them to be
> different directories?
>
> I'll point out that PostgreSQL backups absolutely require the WAL in
> order to be able to be restored (at least the WAL generated during the
> backup) in order to reach consistency. If the two were to be seperated,
> you'd probably still want the backups to 'stand alone' and that would
> mean duplicating all of the WAL which is created during the backup and
> storing it with the backup.
>
> The first question really is why you're looking for this though..? If
> there's a good use-case for it, we could look at adding that it as an
> option.
>
> Thanks!
>
> Stephen
>
From | Date | Subject | |
---|---|---|---|
Next Message | Corey Taylor | 2017-12-16 09:03:04 | How to see index was rejected for seq scan? |
Previous Message | John R Pierce | 2017-12-15 21:18:29 | Re: pgbench |