Re: point in time recovery and moving datafiles online

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Tatsuo Ishii <t-ishii(at)sra(dot)co(dot)jp>
Cc: marc(at)bloodnok(dot)com, pgsql-hackers(at)postgresql(dot)org
Subject: Re: point in time recovery and moving datafiles online
Date: 2002-03-08 02:13:40
Message-ID: 26127.1015553620@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general pgsql-hackers

Tatsuo Ishii <t-ishii(at)sra(dot)co(dot)jp> writes:
>> No. It is not reasonable to expect the backup to allow you to recreate
>> any state occurring before the *end* of the backup process.

> I see your point. But is it reasonable? We cannot know when the backup
> process completes beforehand and that makes the archive log recovery
> less usefull in my opinion.

So? The backup will not let you recover states occurring before it
starts, either. I don't see the issue. Anyone using this in a
production situation will *always* have a prior backup and a continuous
WAL trail since the start of the prior backup. Whether the
synchronization point is the start time or end time of each new backup
isn't a hot issue for them --- at least it shouldn't be if they are
following appropriate procedures.

The only way I can see to make it work the other way is to dump the
entire contents of any to-be-deleted table into the WAL log before
deleting it. That is way too much overhead to be a reasonable tradeoff
for letting people think of the start time rather than end time of their
backups as the cutover point.

regards, tom lane

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Lamar Owen 2002-03-08 02:26:07 Re: Installing 7.2 on RedHat 7.1
Previous Message Tom Lane 2002-03-08 02:03:54 Re: PostgreSQL for PS/2 ?

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2002-03-08 02:21:53 Re: pg_dump doesn't report failure
Previous Message Bruce Momjian 2002-03-08 02:05:12 Re: TODO question