Wait for RPC handlers to drain before checking subscription state #4771
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The Workload API handler tests assert that cache subscriptions are properly closed. To test for this, they wait until the gRPC server is done serving (e.g. GracefulStop) and then check the state of returned subscriptions.
Unfortunately, if a streaming RPC is cancelled on the client side (e.g. context cancellation), it can result in the RPC to "complete" on the server side before the actual handler implementation has returned. This became more apparent with a recent gRPC change (see #4746) which has caused CI/CD failures.
To mitigate this, this change updates the tests to use a stats handler to track how many RPCs are still "open" and waits for all RPCs to close before checking the subscription handler.