Thread

Re: Orphan page in _bt_split

Konstantin Knizhnik <knizhnik@garret.ru> — 2025-09-04T11:44:41Z

On 04/09/2025 3:55 AM, Peter Geoghegan wrote:
> On Wed, Sep 3, 2025 at 2:32 AM Konstantin Knizhnik <knizhnik@garret.ru> wrote:
>> But sooner or later vacuum will be called for this index and will
>> traverse this page, will not it?
>> There is not other way to reuse this page without deleting it or I am
>> missing something?
> That's true. But VACUUM won't even attempt to delete it unless it can
> also remove all of the index tuples. Which, in general, probably won't
> happen (it happened with your test case, but that's probably not
> typical).
>
>> But vacuum is not just logging this message. It throws error which means
>> that vacuum for this relation will be performed any more.
> What error? You showed an assertion failure, but that won't be hit in
> release builds.
>
Sorry, I missed it with another "failed to re-find" error ("... for split").
Yes, this error can happen only for Postgres build with enabled casserts.