From: | Simon Riggs <simon(at)2ndQuadrant(dot)com> |
---|---|
To: | Florian Pflug <fgp(at)phlo(dot)org> |
Cc: | Aidan Van Dyk <aidan(at)highrise(dot)ca>, Daniel Farina <daniel(at)heroku(dot)com>, Chris Redekop <chris(at)replicon(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Hot Backup with rsync fails at pg_clog if under load |
Date: | 2011-10-27 14:30:06 |
Message-ID: | CA+U5nML80NWYAYt0=-ioA1dYSX5++b=JODp9nu0tv4DKDsaM3g@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Thu, Oct 27, 2011 at 3:03 PM, Florian Pflug <fgp(at)phlo(dot)org> wrote:
>> I think you make a good case for doing this.
>>
>> However, I'm concerned that moving LogStandbySnapshot() in a backpatch
>> seems more risky than it's worth. We could easily introduce a new bug
>> into what we would all agree is a complex piece of code. Minimal
>> change seems best in this case.
>
> OTOH, we currently compute oldestActiveXid within LogStandbySnapshot().
> Your proposed patch changes that, which also carries a risk since something
> could depend on these values being in sync. Especially since both the logged
> snapshot and oldestActiveXid influence the snapshot tracking on the slave.
>
> But since you wrote most of that code, your judgement about the relative
> risks of these two approaches obviously out-weights mine.
We must move oldestActiveXid since that is the source of a bug. There
is no need to move LogStandbySnapshot(), so I am suggesting we don't
do that for the backpatch. I was going to implement it the way you
suggest in HEAD, since I agree that is a cleaner way.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2011-10-27 14:54:00 | Re: Updated version of pg_receivexlog |
Previous Message | Magnus Hagander | 2011-10-27 14:14:04 | Re: Your review of pg_receivexlog/pg_basebackup |