Control: reopen -1
Re: Tomas Barton
> Sorry to bring this up again. I've tried to reproduce the issue on a
> postgresql-13 cluster that wasn't affected by this bug.
Oh, that's a real issue then, reopening. I can confirm the crash here:
2021-09-09 14:18:33.448 CEST [1538480] LOG: server process
I've tried to rule out a possible data corruption issue, server is using
ECC RAM modules, checksums are enabled on database. I've tried rebuilding
indexes, running VACUUM FULL, switching to replica etc.
Anyway the issues was probably caused by a faulty NVMe adapter:
kernel: pcieport :00:01.1:
Here's an attempt to make the bug reproducible. Unfortunately I'm not able
to reproduce the issue with generated data.
dropdb testdb || true
createdb -E UTF8 testdb
cat < stress.sql
CREATE TABLE "public".downloaded_images (
itemid text NOT NULL,
property text NOT NULL,
image_number intege
Re: Tomas Barton
> a slightly sophisticated SELECT query with a CURSOR can lead to
> postgresql server segmentation fault.
>
> LOG: server process (PID 10722) was terminated by signal 11: Segmentation
> fault
> DETAIL: Failed process was running: COMMIT
> I'll try to make an reproducable code,
Package: postgresql-13
Version: 13.4-1.pgdg100+1
Severity: normal
Dear Maintainer,
a slightly sophisticated SELECT query with a CURSOR can lead to
postgresql server segmentation fault.
LOG: server process (PID 10722) was terminated by signal 11: Segmentation
fault
DETAIL: Failed process was ru
5 matches
Mail list logo
