Horton

Re: Race condition in recovery?

Kyotaro Horiguchi <horikyota.ntt@gmail.com>

View thread View on postgr.es
From: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
To: robertmhaas@gmail.com
Cc: dilipbalaut@gmail.com, hlinnaka@iki.fi, pgsql-hackers@lists.postgresql.org
Date: 2021-06-07T05:01:45Z
Lists: pgsql-hackers 
Sorry, some extra words are left alone.

At Mon, 07 Jun 2021 13:57:35 +0900 (JST), Kyotaro Horiguchi <horikyota.ntt@gmail.com> wrote in 
> As I said upthread the relationship between receiveTLI and
> recoveryTargetTLI is not confirmed yet at the point.
- findNewestTimeLine() simply searches for the history file with the
- largest timeline id so the returned there's a case where the timeline
+ findNewestTimeLine() simply searches for the history file with the
+ largest timeline id so there's a case where the timeline
> id that the function returns is not a future of the latest checkpoint
> TLI.  I think that the fact that rescanLatestTimeLine() checks the
> relationship is telling us that we need to do the same in the path as
> well.
> 
> In my previous proposal, it is done just after the line the patch
> touches but it can be in the if (fetching_ckpt) branch.

regards.

-- 
Kyotaro Horiguchi
NTT Open Source Software Center

Commits

  1. Further refinement of stuck_on_old_timeline recovery test

  2. Adjust new test case to set wal_keep_segments.

  3. Adjust new test case to set wal_keep_size.

  4. Fix corner case failure of new standby to follow new primary.

  5. Back-port a few PostgresNode.pm methods.

  6. Allow PostgresNode.pm's backup method to accept backup_options.

  7. Add end-to-end testing of pg_basebackup's tar-format output.

  8. Allow PostgresNode.pm tests to wait for catchup

  9. Delay reading timeline history file until it's fetched from master.