| From: | Bruce Momjian <bruce(at)momjian(dot)us> |
|---|---|
| To: | Andres Freund <andres(at)anarazel(dot)de> |
| Cc: | pgsql-hackers(at)postgresql(dot)org, Heikki Linnakangas <hlinnaka(at)iki(dot)fi> |
| Subject: | Re: crash recovery vs partially written WAL |
| Date: | 2020-12-31 19:13:07 |
| Message-ID: | 20201231191307.GJ22199@momjian.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Wed, Dec 30, 2020 at 12:52:46PM -0800, Andres Freund wrote:
> Hi,
>
> A question from a colleague made me wonder if there are scenarios where
> two subsequent crashes could lead to wrong WAL to be applied.
>
> Imagine the following scenario
> [ xlog page 1 ][ xlog page 2 ][ xlog page 3 ][ xlog page 4 ]
> ^flush ^write ^insert
>
> if the machine crashes in this moment, we could end up with a situation
> where page 1, 3, 4 made it out out to disk, but page 2 wasn't.
I don't see any flaw in your logic. Seems we have to zero out all
future WAL files, not just to the end of the current one, or at least
clear xlp_pageaddr on each future page.
--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EnterpriseDB https://enterprisedb.com
The usefulness of a cup is in its emptiness, Bruce Lee
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Victor Yegorov | 2020-12-31 19:14:46 | Re: Deleting older versions in unique indexes to avoid page splits |
| Previous Message | Thomas Munro | 2020-12-31 19:10:55 | Re: pgbench: option delaying queries till connections establishment? |