Horton
Menu

Re: "PANIC: could not open critical system index 2662" - twice

Kirk Wolak <wolakk@gmail.com>

View thread
From: Kirk Wolak <wolakk@gmail.com>
To: Evgeny Morozov <postgresql3@realityexists.net>
Cc: Andres Freund <andres@anarazel.de>, Thomas Munro <thomas.munro@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, PostgreSQL General <pgsql-general@postgresql.org>
Date: 2023-05-16T19:54:02Z
Lists: pgsql-general 
On Tue, May 16, 2023 at 10:20 AM Evgeny Morozov <
postgresql3@realityexists.net> wrote:

> On 9/05/2023 3:32 am, Andres Freund wrote:
> > Attached is a rough prototype of that idea (only using datconnlimit ==
> > -2 for now).
> > I guess we need to move this to -hackers. Perhaps I'll post subsequent
> > versions below
> >
> https://www.postgresql.org/message-id/20230314174521.74jl6ffqsee5mtug%40awork3.anarazel.de
> ?
> So now that a patch is in the works, can I drop the two corrupted
> databases? Is there a workaround I can use to reduce the risk of running
> into this issue again until a patch is released? (Which I guess would be
> in August?)
>

The only work around to avoid losing data that I know of are backups and
WAL backups.
Plus "hard core testing/validation" that they work.  We settled on
pg_backrest and are happy with it.

Technically, based on what I understand of this bug.  It did not corrupt
the WAL.  If that's true, then if
you had a basebackup and all the wall files, you could have played back and
recovered the data.
At least to some degree.  Assuming I am right.

HTH

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Handle DROP DATABASE getting interrupted

  2. Remove the restriction that the relmap must be 512 bytes.