Skip to content

Conversation

@danolivo
Copy link
Contributor

No description provided.

@danolivo danolivo requested a review from mason-sharp November 12, 2025 08:22
@danolivo danolivo self-assigned this Nov 12, 2025
@danolivo danolivo added the enhancement New feature or request label Nov 12, 2025
4,
0,
INT_MAX,
MAX_KILOBYTES / 1024,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What is reason to change that?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It is just because of boring complain in the log. We assign this GUC in terms of megabytes, so we need to declare units and proper limits too.

@ibrarahmad
Copy link
Contributor

Please resove conflicts

@ibrarahmad ibrarahmad self-requested a review December 4, 2025 11:49
Copy link
Contributor

@ibrarahmad ibrarahmad left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Need to rebase.

@danolivo
Copy link
Contributor Author

danolivo commented Dec 4, 2025

Please resove conflicts

Done.

With this commit, we require that each node of the Spock cluster has only
enabled subscriptions. There is no algorithmic reason to do that, because when
adding the node, we are concerned only about the donor's state and
the consistency of its progress table. However, such a strict procedure ensures
that users can be sure they add a node to a healthy cluster, take action
beforehand if it is not healthy, and that replication doesn't delay WAL cutting.
Introduce a TAP test that could be used to check Z0DAN checks.

XXX: add_node finishes its job with subscriptions in the 'initialising' state.
Do we need to take any action here?

Also, fix the annoying WARNING on the 'exception_replay_queue_size'.
If no group's subscription is enabled, wait_for_sync_event may get stuck in
an infinite loop. Add a check of the subscription state in the waiting loop
to reflect the fact that an apply worker may quietly die, deactivating its
subscription.

Also, add the correct processing of this behaviour to the Z0DAN and
include a TAP test.
Copy link
Contributor

@ibrarahmad ibrarahmad left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@ibrarahmad ibrarahmad merged commit fbe0cbc into main Jan 23, 2026
5 checks passed
@ibrarahmad ibrarahmad deleted the spoc-294 branch January 23, 2026 12:41
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

enhancement New feature or request

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants