| From: | Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com> |
|---|---|
| To: | Simon Riggs <simon(at)2ndQuadrant(dot)com> |
| Cc: | PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Hot standby, recovery infrastructure |
| Date: | 2009-01-27 18:11:52 |
| Message-ID: | 497F4E68.5070308@enterprisedb.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Simon Riggs wrote:
> On Tue, 2009-01-27 at 17:50 +0200, Heikki Linnakangas wrote:
>
>> Hmm, I think we have small issue if the last WAL segment restored from
>> the archive is an incomplete one:
>
> All normal archive recoveries have "complete" WAL files, since an xlog
> switch record jumps past missing entries at the end of the file. So I
> was just about to say "how can it be an incomplete one?", to which the
> answer is file damage of some kind causing a failed CRC check.
>
> So yes, it is possible and no, the patch doesn't work that way yet.
Actually, the code already deals with that possibility. That's the
motivation for the timelines: the new WAL record will be generated with
a new timelineid. We never append to an existing WAL file after archive
recovery.
--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Andrew Sullivan | 2009-01-27 18:15:34 | Re: 8.4 release planning |
| Previous Message | Tom Lane | 2009-01-27 17:52:41 | Re: 8.4 release planning |