| From: | Michael Paquier <michael(dot)paquier(at)gmail(dot)com> |
|---|---|
| To: | David Steele <david(at)pgmasters(dot)net> |
| Cc: | PostgreSQL mailing lists <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Fix checkpoint skip logic on idle systems by tracking LSN progress |
| Date: | 2016-09-28 07:35:52 |
| Message-ID: | CAB7nPqR=TL1aYQn673GBCR+2vTMdbzgNU5xH=80GfAzipJeZ4g@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Wed, Sep 28, 2016 at 6:12 AM, David Steele <david(at)pgmasters(dot)net> wrote:
> I tried the attached patch set and noticed an interesting behavior. With
> archive_timeout=5 whenever I made a change I would get a WAL segment within
> a few seconds as expected then another one would follow a few minutes later.
That's intentional. We may be able to make XLOG_SWITCH records as not
updating the progress LSN, but I wanted to tackle that as a separate
patch once we got the basics done correctly, which is still what I
think this patch is doing. I should have been more precise upthread:
this patch makes the handling of checkpoint skip logic correct for
only standby snapshots, not segment switches, and puts the infra to
handle other things.
--
Michael
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Rushabh Lathia | 2016-09-28 07:57:11 | Re: Showing parallel status in \df+ |
| Previous Message | Michael Paquier | 2016-09-28 07:32:40 | Re: Fix checkpoint skip logic on idle systems by tracking LSN progress |