| From: | Michael Paquier <michael(dot)paquier(at)gmail(dot)com> |
|---|---|
| To: | David Steele <david(at)pgmasters(dot)net> |
| Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Archiving done right |
| Date: | 2015-08-24 02:49:22 |
| Message-ID: | CAB7nPqTuh14sWZ-xhbvtT5y3sCMJQowpoXTPMBqx0hJz_wSMDQ@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Fri, Aug 21, 2015 at 11:25 PM, David Steele <david(at)pgmasters(dot)net> wrote:
> In the current scheme, if the standby has not been archiving and then
> gets promoted, won't it try to push the older WAL segments because the
> .done files do not exist in archive_status? Or does the archiver not
> cross timelines?
The segments are marked with .done on a standby once their reception
is finished at recovery.
If, by current, you mean >= 9.5, it is also mentioned in the docs that
when archive_mode = on and the standby is promoted, the
now-master-standby will not archive segments it did not generate
itself:
http://www.postgresql.org/docs/devel/static/warm-standby.html#CONTINUOUS-ARCHIVING-IN-STANDBY
--
Michael
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Amit Kapila | 2015-08-24 03:43:09 | Re: checkpointer continuous flushing |
| Previous Message | Tom Lane | 2015-08-24 02:29:37 | Re: Performance improvement for joins where outer side is unique |