feat(ha-heartbeat-committed-propagation): migrate last_known_committed_config_tag to _Atomic (task 1.1)

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>
This commit is contained in:
Rongsong Shen 2026-06-20 11:27:46 +00:00
parent 9817813f54
commit f4327ce4f6
6 changed files with 20 additions and 14 deletions

View file

@ -4,7 +4,7 @@
> The work runs along a single request path (atomic migration → libxce sender → primary stamp / standby consume → vote-floor verification → E2E), so tasks are sequential by design; no `(P)` markers. The two foundation tasks (1.1 atomic migration, 1.2 sender) touch independent files but are kept ordered for review simplicity. The HA-critical change is the standby consume + term-guard (task 2.2) and its vote-floor interaction (task 3.2). > The work runs along a single request path (atomic migration → libxce sender → primary stamp / standby consume → vote-floor verification → E2E), so tasks are sequential by design; no `(P)` markers. The two foundation tasks (1.1 atomic migration, 1.2 sender) touch independent files but are kept ordered for review simplicity. The HA-critical change is the standby consume + term-guard (task 2.2) and its vote-floor interaction (task 3.2).
- [ ] 1. Foundation: atomic state + heartbeat sender - [ ] 1. Foundation: atomic state + heartbeat sender
- [ ] 1.1 Migrate `last_known_committed_config_tag` to `_Atomic` - [x] 1.1 Migrate `last_known_committed_config_tag` to `_Atomic`
- Change `ha->last_known_committed_config_tag` from a plain `uint64_t` to `_Atomic uint64_t` (`src/controller/include/ha.h`) and update its existing writers — `cluster_apply_ctrlhello_reply` (`ha_convergence.c`) and the election safe-commit path (`ha.c`) — to `atomic_store_explicit(..., release)`, and any reader (incl. the controller-status `ctrlstatus_fill`) to `atomic_load_explicit(..., acquire)`. Pure type/access migration — no behavior change. - Change `ha->last_known_committed_config_tag` from a plain `uint64_t` to `_Atomic uint64_t` (`src/controller/include/ha.h`) and update its existing writers — `cluster_apply_ctrlhello_reply` (`ha_convergence.c`) and the election safe-commit path (`ha.c`) — to `atomic_store_explicit(..., release)`, and any reader (incl. the controller-status `ctrlstatus_fill`) to `atomic_load_explicit(..., acquire)`. Pure type/access migration — no behavior change.
- Observable: the controller suite builds clean and passes unchanged; every read/write of the field uses atomic accessors (no plain access remains). - Observable: the controller suite builds clean and passes unchanged; every read/write of the field uses atomic accessors (no plain access remains).
- _Requirements: 2.3_ - _Requirements: 2.3_

View file

@ -153,10 +153,12 @@ void ctrlstatus_fill(cluster_t* c, xce_ctrlstatus_t* st)
} else { } else {
st->committedLsn = atomic_load_explicit(&ha->last_known_committed_lsn, st->committedLsn = atomic_load_explicit(&ha->last_known_committed_lsn,
memory_order_acquire); memory_order_acquire);
/* last_known_committed_config_tag is a plain (non-atomic) uint64_t /* last_known_committed_config_tag is an _Atomic uint64_t; an acquire
* written under ha->lock; a best-effort aligned 8-byte read is correct * load mirrors last_known_committed_lsn above and must NOT take the
* here and must NOT take the lock on this read path. */ * lock on this read path. */
st->committedConfigVersion = ha->last_known_committed_config_tag; st->committedConfigVersion =
atomic_load_explicit(&ha->last_known_committed_config_tag,
memory_order_acquire);
} }
st->appliedLsn = atomic_load_explicit(&ha->applied_lsn, st->appliedLsn = atomic_load_explicit(&ha->applied_lsn,
memory_order_acquire); memory_order_acquire);

View file

@ -169,7 +169,7 @@ typedef struct {
void * _Atomic elect_ctx; /* Req 2: ha_elect_ctx_t* during election; NULL otherwise */ void * _Atomic elect_ctx; /* Req 2: ha_elect_ctx_t* during election; NULL otherwise */
_Atomic uint64_t last_known_committed_lsn; /* Req 5: from primary heartbeats */ _Atomic uint64_t last_known_committed_lsn; /* Req 5: from primary heartbeats */
sem_t heartbeat_seen; /* Req 1: posted by heartbeat handler during CTRL_INIT */ sem_t heartbeat_seen; /* Req 1: posted by heartbeat handler during CTRL_INIT */
uint64_t last_known_committed_config_tag; /* Req 8: from primary heartbeats */ _Atomic uint64_t last_known_committed_config_tag; /* Req 8: from primary heartbeats */
_Atomic uint64_t committed_config_tag; /* Req 8.1: set on quorum ACK */ _Atomic uint64_t committed_config_tag; /* Req 8.1: set on quorum ACK */
/* ── New fields added by election-iomgr-refactor spec (Task 2.1) ── */ /* ── New fields added by election-iomgr-refactor spec (Task 2.1) ── */
void *client_ssl_ctx; /* SSL_CTX* (TLS_client_method); created in myapp_run */ void *client_ssl_ctx; /* SSL_CTX* (TLS_client_method); created in myapp_run */
@ -769,9 +769,9 @@ int handle_request_vote(ha_state_t *ha,
* *out_tag = max(local_config_tag, atomic committed_config_tag) * *out_tag = max(local_config_tag, atomic committed_config_tag)
* *out_lsn = max(local_applied_lsn, atomic last_known_committed_lsn) * *out_lsn = max(local_applied_lsn, atomic last_known_committed_lsn)
* *
* Uses acquire loads of the atomic learned-committed fields; never reads the * Uses acquire loads of the atomic learned-committed fields; never reads
* non-atomic last_known_committed_config_tag (data-race safe). Pure aside from * last_known_committed_config_tag (a heartbeat-learned config tag, not the
* the atomic reads — no state mutation. * vote freshness floor). Pure aside from the atomic reads — no state mutation.
*/ */
void ha_vote_self_freshness(ha_state_t *ha, void ha_vote_self_freshness(ha_state_t *ha,
uint64_t local_config_tag, uint64_t local_config_tag,

View file

@ -374,7 +374,8 @@ void cluster_apply_ctrlhello_reply(cluster_t *c, uint64_t committed_lsn,
{ {
atomic_store_explicit(&c->ha.last_known_committed_lsn, committed_lsn, atomic_store_explicit(&c->ha.last_known_committed_lsn, committed_lsn,
memory_order_release); memory_order_release);
c->ha.last_known_committed_config_tag = config_tag; atomic_store_explicit(&c->ha.last_known_committed_config_tag, config_tag,
memory_order_release);
INFO("HA: CtrlHelloReply from primary: committed_lsn=%" PRIu64 INFO("HA: CtrlHelloReply from primary: committed_lsn=%" PRIu64
" config_tag=%" PRIu64, committed_lsn, config_tag); " config_tag=%" PRIu64, committed_lsn, config_tag);

View file

@ -361,7 +361,8 @@ static void test_ctrlstatus_standby_reports_learned_committed(void)
/* learned from the primary's heartbeats */ /* learned from the primary's heartbeats */
atomic_store_explicit(&c.ha.last_known_committed_lsn, 36, atomic_store_explicit(&c.ha.last_known_committed_lsn, 36,
memory_order_release); memory_order_release);
c.ha.last_known_committed_config_tag = 0x00C0FFEE0000CAFEULL; atomic_store_explicit(&c.ha.last_known_committed_config_tag,
0x00C0FFEE0000CAFEULL, memory_order_release);
xce_ctrlstatus_t st; xce_ctrlstatus_t st;
ctrlstatus_fill(&c, &st); ctrlstatus_fill(&c, &st);
@ -393,7 +394,8 @@ static void test_ctrlstatus_standby_prelearn_committed_zero(void)
memory_order_release); memory_order_release);
atomic_store_explicit(&c.ha.last_known_committed_lsn, 0, atomic_store_explicit(&c.ha.last_known_committed_lsn, 0,
memory_order_release); memory_order_release);
c.ha.last_known_committed_config_tag = 0; atomic_store_explicit(&c.ha.last_known_committed_config_tag, 0,
memory_order_release);
/* even if the node-local primary marker had a stale value, off-primary it /* even if the node-local primary marker had a stale value, off-primary it
* must NOT be surfaced — prove the learned (zero) value wins. */ * must NOT be surfaced — prove the learned (zero) value wins. */
atomic_store_explicit(&c.persist.committed_lsn, 77, memory_order_release); atomic_store_explicit(&c.persist.committed_lsn, 77, memory_order_release);
@ -429,7 +431,8 @@ static void test_ctrlstatus_primary_reports_own_committed(void)
/* stale learned values must NOT be surfaced on a primary */ /* stale learned values must NOT be surfaced on a primary */
atomic_store_explicit(&c.ha.last_known_committed_lsn, 9, atomic_store_explicit(&c.ha.last_known_committed_lsn, 9,
memory_order_release); memory_order_release);
c.ha.last_known_committed_config_tag = 0x1111ULL; atomic_store_explicit(&c.ha.last_known_committed_config_tag, 0x1111ULL,
memory_order_release);
xce_ctrlstatus_t st; xce_ctrlstatus_t st;
ctrlstatus_fill(&c, &st); ctrlstatus_fill(&c, &st);

View file

@ -724,7 +724,7 @@ static void test_ctrlhello_reply_intake_stores_and_reconciles(void)
atomic_load(&g_dc.ha.last_known_committed_lsn), atomic_load(&g_dc.ha.last_known_committed_lsn),
"CtrlHelloReply intake must store last_known_committed_lsn (existing)"); "CtrlHelloReply intake must store last_known_committed_lsn (existing)");
TEST_ASSERT_EQUAL_UINT64_MESSAGE(reported_tag, TEST_ASSERT_EQUAL_UINT64_MESSAGE(reported_tag,
g_dc.ha.last_known_committed_config_tag, atomic_load(&g_dc.ha.last_known_committed_config_tag),
"CtrlHelloReply intake must store the config tag (existing)"); "CtrlHelloReply intake must store the config tag (existing)");
/* Assert (intake integration): the intake also TRIGGERED reconciliation — /* Assert (intake integration): the intake also TRIGGERED reconciliation —