| From: | Gregory Stark <stark(at)enterprisedb(dot)com> |
|---|---|
| To: | "Heikki Linnakangas" <heikki(at)enterprisedb(dot)com> |
| Cc: | "Simon Riggs" <simon(at)2ndquadrant(dot)com>, "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us>, "Joshua D(dot) Drake" <jd(at)commandprompt(dot)com>, "Jeff Trout" <threshar(at)threshar(dot)is-a-geek(dot)com>, <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: [GENERAL] Slow PITR restore |
| Date: | 2007-12-13 22:16:39 |
| Message-ID: | 87d4tansu0.fsf@oxford.xeocode.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general pgsql-hackers |
"Heikki Linnakangas" <heikki(at)enterprisedb(dot)com> writes:
> It would be interesting to do something like that to speed up replay of long
> PITR archives, though. You could scan all (or at least far ahead) the WAL
> records, and make note of where there is full page writes for each page.
> Whenever there's a full page write further ahead in the log, you could ignore
> all changes to that page before that, because they're going to be overwritten
> anyway.
Hm, you could generate that data when you generate the logs. Keep a hash of
block number and when the last full page write was. Then whenever you switch
log files dump out that hash in a second file alongside.
PITR recovery could read that in before it starts reading any file and consult
it before applying any records.
--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com
Ask me about EnterpriseDB's PostGIS support!
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Andrew Sullivan | 2007-12-13 22:17:24 | Re: Better alternative for Primary Key then serial?? |
| Previous Message | Tom Lane | 2007-12-13 22:10:44 | Re: [GENERAL] Slow PITR restore |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Josh Berkus | 2007-12-14 00:09:24 | Re: [GENERAL] Slow PITR restore |
| Previous Message | Tom Lane | 2007-12-13 22:10:44 | Re: [GENERAL] Slow PITR restore |