Re: several questions about pg_dumpall, pg_start_backup, pg_basebackup and WAL

From: Patrick Dung <patrick_dkt(at)yahoo(dot)com(dot)hk>
To: Kevin Grittner <kgrittn(at)ymail(dot)com>, Albe Laurenz <laurenz(dot)albe(at)wien(dot)gv(dot)at>, "pgsql-admin(at)postgresql(dot)org" <pgsql-admin(at)postgresql(dot)org>
Subject: Re: several questions about pg_dumpall, pg_start_backup, pg_basebackup and WAL
Date: 2013-09-01 09:33:33
Message-ID: 1378028013.14431.YahooMailNeo@web193502.mail.sg3.yahoo.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin pgsql-hackers

'Crash consistent' is a term I heard in some oracle white papers..
If I am correct, it means the database would be consistent after the crash recovery is done.

>> Which means it is consistent at the time that the pg_start_backup
>> is run.

>No.  It will be consistent with the time that pg_stop_backup was
>run, or any later point in time that you choose, as long as you
>have WAL to that point in time.
In the manual, we where told to do three steps:
1) pg_start_backup

2) backup by tar
3) pg_stop_backup
If the backup is consistent when pg_stop_backup is called, I think it worth to mention it in the manual.

Thanks and regards,
Patrick Dung

________________________________
From: Kevin Grittner <kgrittn(at)ymail(dot)com>
To: Patrick Dung <patrick_dkt(at)yahoo(dot)com(dot)hk>; Albe Laurenz <laurenz(dot)albe(at)wien(dot)gv(dot)at>; "pgsql-admin(at)postgresql(dot)org" <pgsql-admin(at)postgresql(dot)org>
Sent: Wednesday, August 28, 2013 3:31 AM
Subject: Re: [ADMIN] several questions about pg_dumpall, pg_start_backup, pg_basebackup and WAL

Patrick Dung <patrick_dkt(at)yahoo(dot)com(dot)hk> wrote:
> Albe Laurenz <laurenz(dot)albe(at)wien(dot)gv(dot)at>
>> Patrick Dung wrote:

>>> It is possible that there is file changes (added or file size
>>> changed) between the pg_start_backup and pg_stop_backup.
>>
>> Yes.
>>
>>> Is the backup consistent?
>>
>> The tar or snapshot itself will not be consistent, it will have
>> to be recovered at least until the end on the online backup.
>
> I should ask: is the backup crash consistent?

PITR restore procedures will use the crash recovery mechanism to
make the database consistent, if that's what you mean.

> Which means it is consistent at the time that the pg_start_backup
> is run.

No.  It will be consistent with the time that pg_stop_backup was
run, or any later point in time that you choose, as long as you
have WAL to that point in time.

--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

In response to

Browse pgsql-admin by date

  From Date Subject
Next Message boraldomaster 2013-09-04 10:55:51 Re: Postgres 9.1 statistics in pg_stat_database
Previous Message françois Figarola 2013-08-31 05:52:24 Re: to_timestamp error between postgres version 8.3 and 9.2

Browse pgsql-hackers by date

  From Date Subject
Next Message Andres Freund 2013-09-01 11:06:39 Re: INSERT...ON DUPLICATE KEY IGNORE
Previous Message Amit Kapila 2013-09-01 07:37:38 Re: Redesigning checkpoint_segments