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

Update fork #3

Merged
merged 2 commits into from
Jan 6, 2025
Merged

Update fork #3

merged 2 commits into from
Jan 6, 2025

Conversation

pthirun
Copy link
Owner

@pthirun pthirun commented Jan 6, 2025

No description provided.

xunyin8 and others added 2 commits January 2, 2025 16:14
)

Int2ObjectOpenHashMap is not thread safe and using it for ServerReadQuotaUsageStats could result in unexpected behaviors during race conditions since the map is accessed and modified by multiple triggers and threads:

Read traffic
CV/routing change events
Version creation and deletion events
Added a test and confirmed that it will hang/timeout with Int2ObjectOpenHashMap but succeed with a thread safe map like VeniceConcurrentHashMap

Also minor defensive code change in getReadQuotaUsageRatio to avoid NPE since the map could change in between check and access.
…AGG store during L/F transition (#1409)

Ready-to-serve check happens in two threads today:
(1) Drainer: which is reasonable
(2) SIT thread: there are two sub-cases:

During L/F transition: It is only for empty RT topic, and today we have HB for non-agg and A/A already, so it only applies to the lingering AGG system store. In this PR, it is checked against store data replication type, if it is AGG we will still have it. We will completely remove it once AGG mode is removed.
Before subscribe, it has a pre-check for short circuit to complete, and this is fine, as it is before the subscribe.
Also, I noticed that there are a bunch of integration tests which enables incremental push on NON-AA store. This is totally invalid setup and after removing the additional RTS check, these tests start to fail/very flaky.
To make sure we don't do this in test and prod, this PR added a new check in update store command in parent controller, which will fail loudly if the new update request is to set incremental push to be true on a non-A/A store.
I have another thinking about enabling all the configs that are enabled in production to be enabled in the test suite, but I'd like to do it in another separate PR, so it is easier for reviewers.

Beyonds that, fixed a flaky unit test that can throw NPE.
@pthirun pthirun merged commit d45d106 into pthirun:main Jan 6, 2025
13 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.

3 participants