WAL-byte decode proves REAL data loss (not commit-index): committed prefix
entries physically absent from live nodes (ctrl-2 WAL starts LSN 16, ctrl-3 LSN
13), present only on the down node. Mechanism: empty-rejoin catch-up streams
[watermark..committed] not [1..], no snapshot to backfill the prefix -> permanent
gap. Predates libxce-failover. Full findings + open forks + xce_test waldump
tooling plan in INVESTIGATION-failover-job-loss.md. Not yet fixed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Test-only: exercises the real ha_vote_self_freshness/handle_request_vote gate.
A heartbeat raise via ha_observe_heartbeat_committed lifts the floor to
max(durable=5, raised=9); candidate at 7 denied, at 9/12 granted; a lower
heartbeat never lowers it. Discriminating constants fail under either regression
(floor->durable-only or fold->min). No production/vote-path change (Req 3.2).
123/0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
primary_heartbeat_fn now sends xce_ha_heartbeat_req with the primary's OWN
atomic ha->term / persist.committed_lsn / committed_config_tag (NOT last_known_*),
snapshotted once per tick (Req 1.1, 3.4). XCE_TESTING seam adds a capturing
ha_heartbeat_req_hook (precedence new->legacy->real) so existing ha_nop_req_hook
liveness tests stay green; send-failure WARN + last_push_time unchanged. New test
poisons last_known_* to prove own-committed. Direct binary 119/0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Append a committed-consume branch to the OP_nop handler (existing liveness/reply
byte-unchanged): ha_consume_heartbeat_committed applies the non-primary gate
(3.4) + stale-term guard (accept hb->term >= ha->term, 3.2), then delegates to
the monotonic ha_observe_heartbeat_committed (1.2). Absent body / zero / failed
gate are no-ops (4.2). No vote-floor edit — safety is the term-guard. 4 tests.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Strict-monotonic updater for a standby's learned committed position: advance
last_known_committed_lsn only when strictly greater (zero/lower no-op), set the
config tag together (tag-before-lsn release ordering for reader safety). No
ha_rejoin_reconcile, no truncation, no other mutation (Req 2.1-2.3, 3.1, 3.3).
4 revert-detecting unit tests; controller suite 0 failures.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Send the HA heartbeat as OP_nop + the existing HeartbeatMsg body carrying
{term, committed_lsn, committed_config_tag} (senderPriority unused). Generic
xce_nop_req untouched; C-side body support pre-existed (no typemap edit). Req 1.1
producer half. Round-trip test proves the 3 fields survive the wire; lib suite
0 failures under ASAN.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Mirror the sibling last_known_committed_lsn: all writes release, reads acquire.
Prereq for a heartbeat-receive writer + the ctrlstatus reader to interact
safely (Req 2.3). Pure access migration, no behavior change; controller suite
0 failures.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- ha-heartbeat-committed-propagation/brief.md (NEW): the live-fresh off-primary
committed follow-up controller-status Req 10 routed out of boundary. Primary
stamps committed_lsn/config_tag into each heartbeat (HeartbeatMsg fields exist
but are unconsumed); standby monotonically refreshes last_known_committed_*.
- ha-convergence-recovery/research.md: live 'cluster won't hold a primary'
finding (2026-06-20) as a candidate manifestation for the deferred Task 8 E2E,
with a differential (demotion self-restart churn vs prevote thrash vs fsync env)
and a diagnosis-first recommendation. Code is complete; this is runtime evidence.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add ctrlstatus_fmt.h (ctrlstatus_format_time_ms): renders responder_time_ms as
local YYYY-MM-DD HH:MM:SS,mmm, byte-equivalent to the controller log prefix
(tlog/my_tlog_format) so a captured ctrlstatus lines up against ctrl-N.log.
cmd_ctrlstatus prints a responderTime line; TZ-pinned exact-string unit test.
Req 11.2, 11.3. ctrlstatus group passes.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Role-branch the committed sourcing: PRIMARY reports its own authoritative
persist.committed_lsn / committed_config_tag; off-primary reports the LEARNED
ha->last_known_committed_lsn / last_known_committed_config_tag instead of the
primary-only raw markers that read 0 on a standby (fixes the applied>committed
inversion). Stamp responder_time_ms from CLOCK_REALTIME for log correlation.
Read-only (Req 10.1-10.4, 11.1). 4 new TDD tests; controller suite 110/0/0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Append-only UInt64 epoch-millis field for the responder snapshot timestamp
(Req 11.1, wire-format setup). Hand-authored xce_ctrlstatus_t typemap gains a
matching responder_time_ms member; codec round-trips it. Controller suite 1/1 OK.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
EARS requirements for a controller status query (XCE lsid analog): primary identity
(unknown when unelected), answered-by node+role (distinct from cluster view),
configured peers+count, committed config version + content checksum (catch same-version
divergence), term/commit-apply/fenced/cluster-name, failover-aware delivery, test-client
command, and wait-primary switching its probe to it. C tier only.
Conflation fix: gate every kill + durability assertion on cluster health (3 nodes
up + serving + rejoin settle); assert_no_loss probes serveability first and reports
INCONCLUSIVE (not DATA LOSS) when unserveable; confirm-settle re-check distinguishes
catch-up lag from persistent loss; post-kill election SLA configurable (60s default).
DATA LOSS now reported ONLY for a job persistently absent from a SERVING cluster.
The Task 9 tests were marked [x] but were stub/absent. Convert the
correctness-critical ones to real white-box tests, registered as meson tests
under --suite scheduler (leaks-off per convention):
- test_handlers.c (new) — #include ../src/handler.c; covers 9.2/9.3: each
notify/dataset handler enqueues the correct op / skips NULL-empty / -EINVAL.
Asserts the 5.6 fix (update_job -> UPDATE_JOB_FIELDS, not bulk UPDATE_JOB).
- test_job_update.c — add 9.5 (real sched_update_existing_queue, replacing the
former local stub); already covers 9.4 (job) and 9.7 (rusage).
- meson.build — build + register all three scheduler test executables.
Suite GREEN: sched_test 15/15, sched_job_update_test 10/10,
sched_handlers_test 6/6; ASAN-clean.
Honest remaining: 9.6 dequeue worker-resvals APPLY is PARTIAL (enqueue tested;
apply is a dormant path, R-3); 9.8 is a documented compile-time absence check.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The spec/design placed the scheduler-op name lookup as xce_sched_op_name in
lib/helper, but xce_sched_op_t is a scheduler-only type and src/lib must not
depend on scheduler headers — so it is correctly a static sched_op_name() in
scheduler.c. Code kept as-is (correct); docs amended to match:
- Task 1 intro + 1.1/1.3 + 5.7 + 9.1: scheduler-layer placement + layering note.
- design.md Component 8/9, file table, risks: same reconciliation.
Only xce_notify_event_name (int-typed, lib-level) lives in lib/helper.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
