Skip to content

Distributed messaging MSG-005 drops initial message #11

@dirvine

Description

@dirvine

Observed in local cluster: tests/distributed/scripts/run-tests.sh messaging --local-cluster intermittently fails MSG-005 (missing Alice's first message).\n\nNotes:\n- Transport is QUIC/UDP (ant-quic) and per-peer send serialization was added in saorsa-gossip.\n- join_entity now waits (timeout) and sync requests are sent via pubsub + direct membership.\n- Still seeing missing initial message in MSG-005 (often Alice/Bob mismatch).\n\nSuspected causes:\n- race between join_entity completion, pubsub peer initialization, and first publish.\n- sync request timing vs initial broadcast.\n\nNext steps:\n- add deterministic reproduction with tighter tracing around join/publish + peer init.\n- verify pubsub eager peers are updated before first message publish.\n- consider ack/anti-entropy for first message after join.\n

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions