Closed japinli closed 7 months ago
awesome, thanks for the PR!
On Wed, 31 Jan 2024 at 00:45, Jerry Sievert @.***> wrote:
@JerrySievert commented on this pull request.
can you please add a test (in columnar/src/test/regress/sql) for this? I am happy to help you if you need.
Yeah, I will try to add a test. However, when I try to run make check using dd0ef0720, it fails. It seems the first break commit is 9cf098a26.
Here is my environment: $ lsb_release -a No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 22.04.3 LTS Release: 22.04 Codename: jammy
$ pg_config --version PostgreSQL 16.1
$ gcc --version gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 Copyright (C) 2021 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
Here is the regressions test out.
dd0ef0720-regression.diffs.txt 9cf098a26-regression.diffs.txt
@JerrySievert I try to add a test for this crash, however, I cannot run make check locally. Please take a look! Thanks in advance!
Here is the regressions test out.
dd0ef0720-regression.diffs.txt 9cf098a26-regression.diffs.txt
Aha, there is an OOM error on my dev. :(
Hi @japinli, thank you for report and PR - it looks correct to me.
I don't see crash with regression on PG15/PG16.
Hi @japinli, thank you for report and PR - it looks correct to me.
I don't see crash with regression on PG15/PG16.
Hi, @mkaruza, thanks for the review.
Yeah, it doesn't crash. The regression fails because there is an out-of-memory, which leads Postgres process being killed.
It seems 8GB RAM isn't enough for running a regression test. I'm not sure about this.
When
chunk_group_row_limit
is bigger than 110000, there is a crash caused byReadStripeNextVector()
.For example:
PostgreSQL version 16.1, hydra columnar dd0ef07209dd.