The test native_stream5_hostile_bytes_disconnect_with_traceable_reasons is currently flaky in the PR unit-test lane and has been temporarily ignored.
Observed failure:
- workflow: Unit Tests
- job: stable on ubuntu-latest
- failing test:
zebra-network::zakura::testkit::cluster::tests::native_stream5_hostile_bytes_disconnect_with_traceable_reasons
- error:
WaitError { description: "native stream-5 oversize trace", timeout: 5s }
The failure happened after the hostile peer sent an oversized stream-5 frame. The test waits for a ratelimit JSONL row with event = "frame.oversize" and stream_kind = "header_sync", but under CI load that trace row was not observed before the fixed 5 second deadline.
Follow up:
- Determine whether the trace row is delayed, dropped, or the peer disconnect path is racing the trace writer.
- Re-enable the test once it is event-driven or uses a deterministic synchronization point instead of a short fixed trace-read timeout.
The test
native_stream5_hostile_bytes_disconnect_with_traceable_reasonsis currently flaky in the PR unit-test lane and has been temporarily ignored.Observed failure:
zebra-network::zakura::testkit::cluster::tests::native_stream5_hostile_bytes_disconnect_with_traceable_reasonsWaitError { description: "native stream-5 oversize trace", timeout: 5s }The failure happened after the hostile peer sent an oversized stream-5 frame. The test waits for a
ratelimitJSONL row withevent = "frame.oversize"andstream_kind = "header_sync", but under CI load that trace row was not observed before the fixed 5 second deadline.Follow up: