All required tasks are done and feature-level GO is recorded (d19e9f27);
only optional task 8.5 (observing transient catching-up live) remains.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
bug-1879: on a fresh HA cluster the config-holding ctrl-1 started its
term-1 election before its peers ran; connect failures count as "no"
votes (bug-1423), so it lost at once and became a follower with no
leader. ctrl-2/ctrl-3 hold no config and are voters only (by design),
so nothing retried until ctrl-1's 10 s standby liveness timeout: every
cold start had no primary for ~10 s.
- ha_note_leaderless() marks a config-holding node that becomes a
follower with no leader known; the liveness poll then retries the
election (PreVote) after HA_LEADERLESS_RETRY_MS = 1 s, doubling up to
heartbeat_timeout_ms. Primary contact or becoming leader clears it.
- ha_heartbeat_is_fresh() answers "no" while leaderless: last_activity
then only holds the watchdog's arm time, and the failover gates took
it for a live primary.
Live cold start: primary 1.7 s after the first node started (was ~10 s).
bug-1878: ctrlstatus_fill() reported ha->applied_lsn for every role, a
standby-only counter a primary never advances, so every primary showed
"appliedLsn: 0". A primary now reports persist->last_applied_lsn. The
controller-status design doc's "already correct in every role" note is
corrected.
Tests: 5 new in test_ha_failover_prevote.c, 1 new in test_ctrlstatus.c
(RED: expected 4900, was 0); controller suite 1256/0. HA soaks:
sequential 6 rounds PASS, chaotic 6 rounds PASS (zero divergence, zero
job loss).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
- buglog: gitea_issue / gitea_synced_status for the 20 bugs created and the
30 already-synced entries (#25/#35 reopened as partial)
- .kiro/specs/*/spec.json: "gitea" milestone/issue mapping for all 66 specs
(65 milestones + 195 issues created, 158 closed at creation)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Found live during task 8.5's optional catching-up-observation attempt:
a killed-then-reconnected standby can end up occupying a NEW
replicator_t.standbys[] slot while its OLD slot (host/port still
populated, ep == NULL) also matches the same configured peer.
ctrlstatus_fill's primary-side peer-status loop stopped at the first
host:port match, so it permanently reported that peer dead even
minutes after it had genuinely reconnected and fully caught up --
violating Requirement 13.3's "true, currently-known connection status"
guarantee.
Fix: scan all MAX_STANDBYS slots for a given peer and prefer any slot
with ep != NULL over a slot with ep == NULL; fall back to dead only
when no matching slot is live. Read-only, lock scope unchanged from
task 8.2.
TDD RED->GREEN (1216 Tests: 2 Failures -> 0), independently reproduced
by the reviewer (exact RED replay) plus two independent live
kill/backlog/restart/poll cycles on separate fresh clusters, both
confirming the previously-stuck standby now converges to ok.
design.md's Locked Decision 3 corrected to document the fix.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Two independent live 3-node-cluster runs both confirmed the full
Requirement 13 scenario (healthy all-ok, dead-on-kill, unknown-from-
standby-view) with zero bugs found. No source changes -- pure
validation task.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Adds ctrlstatus_peer_status_str (mirrors ctrlstatus_role_str's
convention: 0=unknown/1=ok/2=catching-up/3=dead, safe default fallback
for any other code) and extends cmd_ctrlstatus's per-controller print
line to render it alongside each configured controller's existing
name:address, completing the CtrlPeer.status wire field added in task
8.1 and sourced in task 8.2.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Populates CtrlPeer.status (added in task 8.1) inside ctrlstatus_fill's
peer-building loop: self always reports ok; a primary responder sources
every other peer from the live replication connection registry
(replicator_t.standbys[], host+port matched, reusing
repl_active_standby_count's ep!=NULL && !in_catch_up predicate) --
dead/catching-up/ok; a non-primary responder reports unknown for every
non-self peer. A self-identity mismatch (pre-existing misconfiguration)
logs one WARN and falls through to the normal non-self sourcing rather
than forcing any peer to ok. Read-only: a bounded lock-protected scan,
no state mutation.
TDD: 3 new tests (primary-sourced-from-registry, standby-unknown-except-
self, self-mismatch-falls-through) plus an extended no-mutation
assertion; RED->GREEN independently reproduced by the reviewer.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Adds CtrlPeer.status (UInt8, next free ordinal @3 after port) plus the
matching xce_ctrlpeer_t.status typemap field, verified field-for-field
against the regenerated codec with a genuine encode/decode round-trip
test. No sourcing/rendering logic yet (task 8.2/8.3) -- field defaults
to 0 (unknown) until wired.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Second design-review pass caught a factual error: the Constraints
section claimed ctrlstatus_fill "already takes ha->lock" for the
primary-identity read, citing the ORIGINAL Req 1-9 design text rather
than the as-built code -- after the ade0081 fix, that read takes no
lock at all. Corrected: this delta's replicator_t.lock acquisition is
the first lock ctrlstatus_fill has taken since that fix, not a repeat
of an existing in-function pattern (still safe, per the existing
repl_active_standby_count precedent elsewhere in the codebase).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Adds requirements, design, and research for showing each configured
controller's connection status (ok/catching-up/dead/unknown) in
ctrlstatus's peer list, with the same off-primary fidelity pattern
already established (primary reports the true picture; a standby
reports unknown for what it cannot itself observe).
Design review found and fixed one gap before implementation: no
defined fallback if the responder's own name doesn't match any
configured peer entry (a pre-existing misconfiguration class, not new
to this delta) -- now an explicit, logged (WARN) degradation instead
of a silently wrong status.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
/kiro-validate-impl: full suite green (controller 1210/0, lib 368/0,
harness 173/0), live 3-node smoke boot confirms bug-1818's fix holds,
12/12 requirement groups covered, cross-task integration traced at the
source level, dependency direction clean, zero blocked tasks. Prior
2026-07-17 validate_impl record preserved under prior_validate_impl.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Feature-level /kiro-validate-impl found the delta's File Structure Plan
omitted 5 as-built files task 7.3 actually needed (request.h/request.c
-- the real heartbeat_msg_t struct-builder -- plus 3 cascading
signature-conformance test files). Documents the as-built reality and
the already-fixed cmd_ctrlstatus gate text; also notes the pre-existing,
deliberate lack of a checked-in unit test for cmd_ctrlstatus's print
layer (the class of gap bug-1818 exploited) as a known, accepted scope
boundary for future readers.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
phase -> implementation-complete; delta.impl_status records tasks
7.1-7.6 done, including the bug-1818 find-and-fix from live E2E.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Requirement 12 delta (primary election term/time) is now fully
implemented: tasks 7.1-7.6 all complete, including the live 3-node
failover validation that found and confirmed the fix for bug-1818.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Live 3-node E2E validation (task 7.6) found that cmd_ctrlstatus's
electedTerm/electedAt lines (task 7.5) gated on NULLSTR(primaryName),
which is unconditionally empty on every non-primary responder in this
build (ctrlstatus_fill only self-reports the primary's own name --
ha->primary_host is vestigial, a pre-existing Req 2 behavior). This made
electedTerm/electedAt print "unknown" on every standby, forever, even
though elected_at_ms was correctly, non-zero, and convergently populated
the whole time (term/committedLsn/appliedLsn, sourced the same
unconditional way, DID correctly display and converge on the same query).
Fix: gate on status->elected_at_ms == 0 instead -- the field stays
exactly 0 until a real promotion stamps it (either road) or a heartbeat
carries a learned value from an actual primary, unlike term (which can
legitimately bump from mere vote/candidacy activity before any primary
is elected). ctrlstatus_fill itself is untouched; this is purely a
cmd_ctrlstatus display-layer correction.
Also corrects design.md's "Design Delta -- Requirement 12" Locked
Decision 3 (and its dependent traceability/service-contract text) to
document this live-E2E finding, mirroring the spec's existing delta
convention (Req 10/11's own live-E2E correction).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
cmd_ctrlstatus (src/harness/test.c) now prints electedTerm/electedAt
right after the Primary: line, gated on the same NULLSTR(primaryName)
"unknown" check that line already uses. electedTerm prints the existing
term field verbatim (no new wire field, per design's Locked Decision 1);
electedAt reuses the existing ctrlstatus_format_time_ms formatter
already used for responderTime, falling back to "unknown" on a NULL
format result. No leakage of a stale term/elected_at_ms when no primary
is known.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
ctrlstatus_fill now unconditionally reads ha->last_known_elected_at_ms
into the reply's elected_at_ms field -- one atomic-acquire load, no
role-branch, mirroring how the handler already reports raw term
unconditionally. Serves both a self-stamped primary and a
heartbeat-learned standby from the same field. "Unknown" display
gating is deliberately left to the client command (task 7.5), not
this handler.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
The primary echoes its own last_known_elected_at_ms on every outgoing
heartbeat (threaded through xce_ha_heartbeat_req's existing scalar-arg
plumbing, alongside term/committed_lsn/committed_config_tag); a standby
processing a received heartbeat (ha_consume_heartbeat_committed, the sole
production call site) stores it unconditionally into its own
last_known_elected_at_ms, reconcile-free -- no other HA field moves.
ha_observe_heartbeat_committed itself is untouched; the new store sits
next to its call site, not inside it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Stamps ha_state_t.last_known_elected_at_ms (CLOCK_REALTIME, epoch-ms) in
both ha_do_promote (non-raft road) and raft_cb_become_leader (raft road),
placed at each road's genuine, final commit point to CTRL_PRIMARY -- past
every early-return refusal path in raft_cb_become_leader (first-boot
phase2 failure, the Live Apply Resource Fence re-check, the promotion
eligibility gate, config-serve-fenced, upgrade-materialization-fenced).
Purely observational: does not change whether/when a promotion succeeds
or is refused. Three new tests (2 positive, 1 negative) prove placement:
a refused promotion leaves the field at its pre-attempt value.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Adds CtrlStatus.electedAtMs@14 and HeartbeatMsg.electedAtMs@4 to the
capnp schema, their hand-authored typemap.h counterparts, and a new
ha_state_t.last_known_elected_at_ms field alongside last_known_committed_lsn.
Schema/state plumbing only -- nothing stamps or consumes these fields yet
(tasks 7.2-7.5). Codec round-trip tests confirm both new fields survive
encode/decode unchanged.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
Adds requirements, design, research, and tasks for reporting the term
and wall-clock time the current primary was elected, including
standby-side learned fidelity (mirroring the Req 10 pattern), following
the existing controller-status spec's delta convention (Req 10/11).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLSNJhNoULvVcUhXEFafQC
All tasks (1.1-6.4) were already [x] with no prior validate_impl
record. Fixed stale delta.impl_status (still said task 6.4 blocked on
sandbox overlayfs despite tasks.md recording it VERIFIED live since
2026-07-02).
Full suite 9/9 green, 36/36 requirement acceptance criteria mapped,
cross-task wire contract verified field-for-field end to end (schema
-> typemap -> handler -> dispatch -> libxce client -> CLI), two live
smoke runs (3-round sequential HA soak + targeted 3-node deploy
exercising Req 10/11 directly) both PASS. One transient test failure
during the run traced to a concurrent second session competing for
the same build directory, not a regression -- reproduced clean 3x in
isolation.
controller-status, ha-commit-engine, ha-heartbeat-committed-propagation,
ha-prevote, ha-single-leader-role, ha-wal-replication, libxce-failover, and
raft-ha-core-safety each had every numbered sub-task marked [x] but their
group-header lines (e.g. "1. Foundation: ...") left unchecked -- a cosmetic
gap, confirmed by roadmap.md's own IMPLEMENTED/VERIFIED notes and zero real
unchecked sub-tasks in any of the eight files. No task content changed.
ha-commit-durability: closes the "pending confirmation under simultaneous
total outage" caveat. Reproduced cold-restart-all (kill all 3 members +
restart) 5 times against a live 3-node cluster on genuine ext4 (not the
sandbox's overlayfs) — zero committed-job loss across 40 rolling-failover
rounds + 5 cold-restart-all cycles. Confirms bug-795 was an overlayfs
fdatasync artifact, not a real durability gap; no production fix needed.
Commit Durability is now proven for both rolling single-node failover and
simultaneous total-cluster outage (src/controller/CLAUDE.md updated).
Live-validated and marked complete against a real 3-node cluster:
- controller-status 6.4: off-primary committed fidelity + responder
timestamp, confirmed via a real failover
- ha-heartbeat-committed-propagation 4.1: standby committedLsn tracks the
primary via heartbeat within the 3s interval
- ha-wal-replication 5.4: standby WAL replication verified via committedLsn
match (job-facing ops are correctly primary-only by design; standbys
return E_XCE_STANDBY, not a bug)
- libxce-failover 6.2: live 3-node failover E2E, jobsub raced against a
primary kill resolves to exactly one job, no duplication
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SDWVNQLKxqTVHh42cH7rV3
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.
