> 23 мая 2022 г., в 10:40, Michael Paquier <michael@paquier.xyz> написал(а):
>
> It took me some time to write a script to bisect that, but I have been
> able to establish a correlation with d9d0762 that causes VACUUM to
> ignore transactions doing some concurrent reindex operations.
I've transformed Peter's test into TAP test that runs ~20 seconds and reliably reproduces problem on my laptop.
And I observe that commenting out condition in following code fixes the test.
//if (!(statusFlags & PROC_IN_SAFE_IC))
h->data_oldest_nonremovable =
TransactionIdOlder(h->data_oldest_nonremovable, xmin);
Best regards, Andrey Borodin.