Root cause (found via ha-convergence-recovery task 8 live 3-node E2E): a
controller with no local config (config_version_tag==0) could become PRIMARY
and push an empty config, breaking replication.
Two upstream election-safety gaps, both outside ha-convergence-recovery:
1. RDHUP re-election bypassed the no-config 'voter only' guard. The startup
path (ha_start_election_async) refuses to start an election when
my_config_tag==0 && n_peers>1, but the RDHUP failover path
(ha_on_primary_ctrl_fd_rdhup) called raft_on_leader_lost unconditionally.
Fix: mirror the guard before elect_ctx alloc / hook / raft_on_leader_lost;
a config-less node stays CTRL_STANDBY (voter only) and grants the
configured peer instead of starting (and winning) its own election.
2. Voter freshness gate compared the candidate only against the voter's LOCAL
config, ignoring the cluster's known committed state. Fix: ha_vote_self_freshness
floors the comparison at max(local, learned-committed config_tag / committed_lsn),
so a voter denies any candidate whose config/cluster state is outdated vs
what the cluster has committed. Strictly additive (never denies the freshest
node — Leader Completeness preserves liveness).
TDD: test_rdhup_noconfig_node_does_not_elect + configured-node contrast;
test_ha_vote_gate learned-floor deny + liveness + backward-compat. Full
controller suite green under ASAN. commit.c and ha-convergence-recovery
files untouched.
configAck branch: per-push majority (acked_config_lsn >= push_tag,
1+count > n_peers/2 over r->ce->n_peers, n_peers<=1 immediate,
ce==NULL -EINVAL); legacy >0 counting and n_active==0 shortcut dead.
push_config_to_standbys returns the barrier result; upgrade failure ->
send_upgrade_error (rc forwarded), no success ACK, no restart,
committed_config_tag success-only.
Remediation r1 (two CRITICAL review findings):
- F1: wire production config-ACK ingestion — replconfigpushreplymsg
decode branch in repl_ack_decode_and_dispatch -> repl_on_config_ack
on the promoted-vote ep; + defensive cluster.c OPCode_reply case.
Closes the unlanded half of ha-config-bootstrap 8.1; without this the
barrier would brick every multi-member config upgrade. No schema change.
- F2: keep the barrier off the iomgr thread — export
push_config_transport_only (non-blocking) for the 5 ha.c + join.c
streaming sites; barrier path now vtask-only (do_upgrade_config).
Wire-path test drives real capnp ACK bytes through the production decode
branch. 30/30 rollout, 6/6 repl_majority, 46/46 config_repl; census 0
FAIL; pre-existing fixture leak only (less than baseline).
notify/update/setpreason/clrpreason admitted in CTRL_PRIMARY without
write quorum (Req 3.2 / umbrella 2.8 / bug-396): all four terminate in
receipt-style RETAIN fact paths — reviewer-traced no-block proof (no
admitted opcode reaches a barrier/WAIT path; request opcodes stay
rejected). 21/21 group; census 494/494 zero FAIL; RED→GREEN
independently reproduced via stash-rebuild.
repl_active_standby_count: registry truth (ep != NULL && !in_catch_up)
under r->lock. ha_state_t gains repl back-pointer (wired at init by
task 5.3); ha_has_write_quorum uses registry truth when wired, exact
legacy fallback when NULL; formula unchanged. Test 29 (wired gate,
catch-up-only standby → NOT_READY, admits after clear) captured RED
pre-fix. 29/29 group tests; suite at the 3-failure baseline.
Handle incoming WalAckReply in repl_ack_decode_and_dispatch: extract
lsn+success, call repl_on_ack on success to update standby acked_lsn
and signal majority-ack waiter. NACK is WARN-logged with no state
change. Two unit tests verify positive path (acked_lsn updated, waiter
unblocked) and negative path (NACK leaves acked_lsn=0, wait times out).
Standby sends WalAckReply via xce_reply after each successfully
applied WAL entry. ha_apply_wal_catchup_entry signature extended
with endpoint_t* and cluster_t* parameters. walackreplymsg union
arm added to RequestReply capnp schema and xce_reply_t. Error
entries skip ACK. Dispatch gate test confirms OPCode_walCatchupEntry
allowed in CTRL_STANDBY/ELECTING/INIT.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Election protocol fixes validated against raft.tla invariants:
- ha_init_raft(): one-time raft state init at startup (re-init would lose
durable election progress); n_peers = full cluster size including self
- handle_request_vote(): higher-term step-down moved BEFORE the role gate
(split-brain fix); standby vote-deny now requires a live primary control
channel (primary_ctrl_fd >= 0) AND a fresh heartbeat
- ha_heartbeat_is_fresh(): derive freshness from ha->last_activity (the
field production actually writes); the orphaned last_heartbeat timespec
had no production writer, leaving the vote-deny gate inert (bug-337);
field removed from ha_state_t
- handle_request_vote(): restore same-candidate idempotent re-grant —
deny only a DIFFERENT candidate in the same term, matching raft.c and
Raft §5.2 votedFor ∈ {Nil, candidateId} (bug-338)
- raft_init(): restore voted_for_term alongside a persisted vote; leaving
it 0 allowed a double vote in the same term across a restart (bug-339)
- myapp_run(): leader_known check treats a vote recorded under this node's
own name as a self-vote — production persists the real candidate name,
the "self" placeholder only appears in test mode
- raft_on_vote_request(): leader steps down on higher-term vote request
instead of denying it
Tests: test_proto_H6_restored_vote_blocks_second_vote added; H1 restore
assertion strengthened; test_ha_vote_gate.c driven through last_activity +
primary_ctrl_fd (the real production fields). Full /xce-build all green
(lib + controller + scheduler + Go suites, 0 failures, ASAN clean).
deploy-ha.sh: start.sh now forwards extra args to the controller binary.
Root cause: task_cmp in delay.c compared nextrun (always 0 for ms-tasks)
instead of nextrun_ts, making all ms-tasks appear equal in the ordered
queue. The heartbeat task was re-enqueued correctly but stuck behind the
liveness poll task in the broken ordering.
Changes:
- delay.c: fix task_cmp to compare nextrun_ts for ms-tasks (ns precision)
- ha.c: use runcount=-1 (infinite repeat) instead of one-shot+re-arm;
simpler and avoids the broken re-arm path entirely
- ha.c: add static atomic idempotency guard to prevent duplicate heartbeat
tasks on re-election
- ha_state_t.n_peers now matches raft_config_t.n_peers convention:
total cluster size (including self), consistently across both layers
- ha_start_election_async: ha->n_peers = n_peers (candidates_count)
- Convergence/counting loops: i < n_peers - 1 (skip self for standby arrays)
- Single-node: n_peers=1, threshold=0 (same behavior, clear semantics)
- ha_elect_ctx_t.n_peers unchanged (separate field for vote array sizing)
- Tests: 3-node n_peers=2->3, single-node 0->1, loop bounds updated
E2E test 6.7: 3-node cluster lifecycle verified
- Initial: ctrl-1 PRIMARY via majority convergence, no degraded mode
- Kill primary: RDHUP detected on both standbys, ctrl-2 elected at term=4
- No data loss: config tag preserved across failover
- Full ASAN build green
- Fix ha_start_election_async: ha->n_peers = n_other (other peers),
not candidates_count (total including self). The standby_converged[]
array is indexed by peer index, so the bound must be n_other.
- Fix ha_compute_safe_committed_lsn: return self_lsn for n_peers==0
(prevents WAL truncation on single-node re-election).
- Integration test: 3-node cluster reaches CTRL_PRIMARY via majority
convergence within ~20s; no degraded-mode WARN in logs.
6.3: Fix single-node bug — ha_compute_safe_committed_lsn now returns
self_lsn instead of 0 for n_peers=0 (prevents WAL truncation on
single-node re-election). Add self_lsn/self_tag to ha_elect_ctx_t,
populated unconditionally in ha_start_election_async. All 3 test
scenarios pass under ASAN.
6.4: Add test_snapshot_recv_128kb_two_chunks (128KB snapshot in
2x64KB chunks, verifies size and applied_lsn); 4 other test cases
already existed from tasks 3.5c/3.5d. All ASAN-clean.
- config_repl.c: INFO at WAL catch-up start with start/target LSN (Req 7.1)
- config_repl.c: INFO at snapshot transfer start and complete with
checkpoint_lsn and size (Req 7.3)
- ha.c committing_timeout_fn_for_test: add WARN log counting
n_converged/n_pending before promotion, matching production path (Req 7.5)
- Reqs 7.2 (standby converged INFO) and 7.4 (majority INFO) were
already present from tasks 3.1 and 5.2 respectively
- Change HA_DEFAULT_COMMITTING_TIMEOUT_MS from 10000 to 60000
- committing_timeout_fn: count converged/pending standbys from
standby_converged[], log WARN with counts, then promote (degraded mode)
- Replace hardcoded 10000 in test_ha_evaluate_committing.c with macro
- Three new tests: default=60000, degrade-promote with 1/2 and 0/2 converged
- In raft_cb_become_leader, after persist_truncate_wal_above(safe_lsn),
store safe_lsn to ha->last_known_committed_lsn via atomic release
- Test: 3-node voters [100,150,200] → safe_lsn=150; asserts
ha->last_known_committed_lsn == 150 after raft_cb_become_leader_for_test
- Add ha_repl_send_snapshot_hook seam in ha.c/ha.h for XCE_TESTING
- test_dispatch_repl_action_send_snapshot: verifies SEND_SNAPSHOT
decision triggers hook when standby_lsn < oldest_wal_lsn
- test_snapshot_recv_two_chunks_assembles_and_renames: verifies standby
assembles snap.part across chunks and renames on isLast=true
- test_snapshot_recv_reconnect_discards_snappart: verifies O_TRUNC
semantics on chunk_index==0 restart after reconnect
- All tests pass under ASAN
- SEND_SNAPSHOT uses checkpoint_lsn (not committed_lsn) for file lookup
- ha_send_snapshot_to_standby stats file for actual total_size
- ha_compute_oldest_wal_lsn scans WAL dir for min segment LSN
- Removed hardcoded 131072 and oldest_wal_lsn=0 placeholders
ha_stream_wal_to_standby() reads WAL from committed_lsn range, accumulates
entries to 256/64KB per batch, enqueues all via xce_request() (non-blocking).
Partial final batch sent immediately. Wired into STREAM_WAL case.
3 tests: 500-entries/2-batches, partial batch, empty range.
Fixed stale task-number comments per review.
ha_dispatch_repl_action_for_standby() calls push_config_to_standbys()
when PUSH_CONFIG decision. Hook seam for unit tests. Wired into
repl_ack_decode_and_dispatch after ha_process_standby_state.
oldest_wal_lsn=0 placeholder documented for task 3.4.
ha_evaluate_repl_action() pure function: stale tag→PUSH_CONFIG,
lsn<oldest_wal→SEND_SNAPSHOT, lsn<committed→STREAM_WAL, else CONVERGED.
5 tests. Fixed misleading stream_wal_from_zero test name per review.
OPCode_nop handler now sends stateReport {applied_lsn, config_tag}
instead of nodata. Function in ha.c for unit-test linkability.
Two tests: warm state and zero-state fresh standby (Req 1.4).
Extend ha_apply_wal_catchup_entry with batch branch:
- Check n_entries > 0, iterate entries[] list in order
- Apply each via apply_wal_entry(); skip NULL/errors without crash
- Advance applied_lsn once after full batch (not per-entry)
- n_entries == 0 falls through to task 2.1 single-entry path
4 new unit tests covering full batch, partial batch, error skip, fallback
Replace stub with ha_apply_wal_catchup_entry() calling apply_wal_entry().
CAS-advance applied_lsn on success; log ERROR and skip on failure.
Standby stays in CTRL_STANDBY throughout. 3 new unit tests.
Add standby_reported_tag/lsn/converged arrays to ha_state_t, sized
MAX_HA_CANDIDATES. Zero-init via existing memset in ha_init().
Unit test verifies zero-init and writability.
Call ha_start_primary_heartbeat in raft_cb_become_leader when entering
CTRL_PRIMARY_COMMITTING so standbys receive NOPs during the committing
window. Fixes 60s stability: 0 false RDHUP events on standbys.
Extend primary_heartbeat_fn to read last_push_time per standby and send NOP
only when elapsed >= heartbeat_interval_s. on_standby_connected writes
last_push_time before ctrlhello send. Remove workaround xce_nop_req at end
of on_standby_connected. Add ha_nop_req_hook XCE_TESTING seam. 3 new tests.
Remove hb_vtask_t, hb_timeout_thread, s_hb_vtask from ha.c. Replace with
liveness_poll_ctx_t + delaymgr liveness_poll_fn: checks cancelled flag and
role==CTRL_STANDBY before acting; re-arms at heartbeat_interval_ms/2.
ha_on_primary_ctrl_fd_rdhup calls ha_disarm_heartbeat_timeout_vtask first
to prevent duplicate failover (Req 5.4). ha_on_heartbeat_received writes
last_activity via clock_gettime. Add test_ha_liveness.c with 7 unit tests.
Add time_t last_activity to ha_state_t for standby liveness timestamp tracking.
Add time_t last_push_time to standby_ep_ctx_t for primary idle NOP suppression.
Zero-initialize last_activity in ha_init(); last_push_time zeroed by calloc.
Add test_ha_init_last_activity_is_zero to test_ha_startup.c.
Task 7.1 (unit tests) was already complete with 8 tests in
test_ha_connection_reuse.c. Added two new tests:
Task 7.2 — test_follower_passive_no_outbound_connect (Req 2.1, 2.4):
Verifies raft_cb_become_follower sets role=CTRL_STANDBY and nulls all
votes[i].ep. The old fetch_config_from_primary path is gone; standbys
never initiate outbound connects after election loss.
Task 7.3 — test_promoted_ep_repl_ack_promotes_to_primary (Req 5.2, 5.4):
Injects two OPCode_replCatchUpDone packets via peer_ctrl_msg_handler in
CTRL_PRIMARY_COMMITTING with n_peers=4 (5-node cluster, quorum=2 ACKs).
Verifies repl_ack_count reaches 2 and role transitions to CTRL_PRIMARY.
Added vote_ep_ctx_set_cluster_for_test() seam so vctx->c is non-NULL
when ha_repl_ack_received fires ha_promote_to_primary.
All 62 controller tests pass; ASAN clean.
spec: .kiro/specs/ha-peer-connection-reuse/tasks.md all tasks [x]
- config_repl.c: add handle_repl_config_push — standby handler for
OPCode_replConfigPush. Verifies SHA-256, saves blob to disk via
conf_save_config, reloads via load_primary_config, ACKs primary.
- config_repl.h: declare handle_repl_config_push.
- cluster.c: dispatch OPCode_replConfigPush to handle_repl_config_push;
populate c->myhost_info.name from myconf->name (ctrl-1/ctrl-2/ctrl-3)
instead of the static "controller" placeholder.
- ha.c: fix raft_cb_become_leader push_config_to_standbys call to pass
real config bytes and tag read from disk (was NULL/0 — standbys got
empty push and never saved config).
- typemap.h: add raw_sha256/raw_bytes/raw_bytes_len fields to
repl_config_push_req_t (mirrors upgrade_config_req_t pattern).
- request.c: pre-extract List(UInt8) sha256/bytes from OPCode_replConfigPush
before capn_free() in xce_decode(); free them in xce_free_request().
- helper.c: add all HA/replication opcode names to xce_request_op_name
so "unknown opcode 30" becomes "replConfigPush" in logs.
- cluster_def.h: add embedded myhost_info field to cluster_t struct.
- ha.h: declare ha_start_primary_heartbeat; remove dead
ha_find_and_connect_to_primary declaration.
- ha.c: add raft_cb_trace callback wired to s_real_cbs.on_trace; emits
RAFT[pN] term=T event=E at DEBUG3 level, visible with -vvv flag.
- ha.c: repl_ack_decode_and_dispatch now silently drops OPCode_nop and
OPCode_reply instead of WARNING — heartbeat NOP replies from standbys
arrive on promoted endpoints and are expected.
- deploy-ha.sh: fix --vcpu -> --vcpus; add \$VERBOSE passthrough so
start.sh respects VERBOSE=-vvv from the environment.
