Re: deadlock-hard flakiness

From: Andres Freund <andres(at)anarazel(dot)de>
To: pgsql-hackers(at)postgresql(dot)org, Thomas Munro <thomas(dot)munro(at)gmail(dot)com>, Robert Haas <robertmhaas(at)gmail(dot)com>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Subject: Re: deadlock-hard flakiness
Date: 2023-02-08 22:15:28
Message-ID: 20230208221528.4w7ecvgcua4hbpom@awork3.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

On 2023-02-08 14:11:45 -0800, Andres Freund wrote:
> On 2023-02-07 17:10:21 -0800, Andres Freund wrote:
> I don't really understand how GetSafeSnapshotBlockingPids() can end up finding
> deadlock-hard's sessions being blocked by a safe snapshot. Afaict nothing uses
> serializable in that test. How can SXACT_FLAG_DEFERRABLE_WAITING be set for
> the sxact of a backend that never did serializable? Are we possibly forgetting
> to clear it or such?
>
>
> I don't think it should affect the reports here, but I did break something
> when removing SHMQueue - GetSafeSnapshotBlockingPids() doesn't check
> output_size anymore. Will fix. Thomas, any chance you could do a pass through
> 96003717645 to see if I screwed up other things? I stared a lot at that
> change, but I obviously did miss at least one thing.

Argh, it's actually caused by 96003717645 as well: Previously loop iteration
without finding a matching pid ends with sxact == NULL, now it doesn't.

Greetings,

Andres Freund

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2023-02-08 22:15:59 pgsql: Stop recommending auto-download of DTD files, and indeed disable
Previous Message Andres Freund 2023-02-08 22:11:45 Re: deadlock-hard flakiness