Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Use ordered queries to reduce deadlocks [RHELDST-22078] #655

Merged
merged 1 commit into from
Dec 19, 2023

Conversation

rohanpm
Copy link
Member

@rohanpm rohanpm commented Dec 19, 2023

There are still DB deadlocks occurring from time to time. These all seem to be related to multiple rhsm-pulp repos updating records for the same files concurrently (e.g. multiple repos having the same RPMs will all need to update the same paths under /origin/ around the same time).

While the relevant code is all prepared to retry on deadlock anyway and so this should not be noticeable beyond a delay, it would be nice to fix this.

This commit tries to fix it by always applying a consistent order for all queries obtaining row locks on items. The change is speculative as I can't reproduce the deadlocks on demand, but it's consistent with the advice given in postgres docs[1]:

The best defense against deadlocks is generally to avoid them by being
certain that all applications using a database acquire locks on
multiple objects in a consistent order

[1] https://www.postgresql.org/docs/current/explicit-locking.html#LOCKING-DEADLOCKS

There are still DB deadlocks occurring from time to time. These all seem
to be related to multiple rhsm-pulp repos updating records for the same
files concurrently (e.g. multiple repos having the same RPMs will all
need to update the same paths under /origin/ around the same time).

While the relevant code is all prepared to retry on deadlock anyway and
so this should not be noticeable beyond a delay, it would be nice to fix
this.

This commit tries to fix it by always applying a consistent order for
all queries obtaining row locks on items. The change is speculative as I
can't reproduce the deadlocks on demand, but it's consistent with the
advice given in postgres docs[1]:

> The best defense against deadlocks is generally to avoid them by being
> certain that all applications using a database acquire locks on
> multiple objects in a consistent order

[1] https://www.postgresql.org/docs/current/explicit-locking.html#LOCKING-DEADLOCKS
@rohanpm rohanpm marked this pull request as ready for review December 19, 2023 00:36
@rohanpm rohanpm merged commit 451cb9c into release-engineering:master Dec 19, 2023
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants