---
id: "bug-990-fix"
status: "fixed"
status_updated: "2026-09-24"
status_note: ""
component: "controller"
severity: "unknown"
timestamp: "2026-06-22T16:00:00.000000"
last_seen: "2026-09-19T09:00:00Z"
occurrences: 1
files: ["src/controller/repl.c (repl_broadcast)", "src/controller/persist.c (persist_wal_async call site + entry_term_locked)", "src/controller/config_repl.c (ha_stream_wal_to_standby catch-up)", "src/controller/ha.c (ha_apply_wal_catchup_entry receive route)", "src/lib/request.c (wal_entry_to_wire per-entry prev tuple encode)"]
tags: ["ha-term-safety", "task-3.4", "live-send-retirement", "term-0", "repl_broadcast", "bug-990", "FIXED", "catch-up-term"]
related: ["bug-990", "bug-851"]
legacy: true
---

# bug-990-fix: ha-term-safety 5.2 live failover: standbys persist per-entry WAL term=0 while primary persists the…

## Description
ha-term-safety 5.2 live failover: standbys persist per-entry WAL term=0 while primary persists the real term; clean-failover committed prefix not byte-identical in TERM (R5.1 term-identity gap). Original bug-990.

## Environment
_Not recorded (migrated from buglog.json)._

## Reproduction
_Not recorded (migrated from buglog.json)._

## Root cause
The LIVE replication SEND paths (repl_broadcast live append + ha_stream_wal_to_standby catch-up) built wal_catchup_entry_t via memset(0) and never set entry.term/prev_log_index/prev_log_term, so standbys received term=0 entries and applied them via the LEGACY term-0 direct-apply branch (persist_wal_write_replicated stamps the STANDBY's local p->term, not the leader's). The term-carrying library receive route (raft_on_append_entries_request, task 3.2) only fired for term!=0 entries, which the live send never produced. Additionally wal_entry_to_wire only encoded per-entry term, not the per-entry prev tuple, so multi-entry catch-up batches could not be routed through the library one-at-a-time.

## Fix
Task 3.4 minimal cutover (chosen over driving raft_send_append_entries live, which would rip out the repl model and risk 848/849/850): (1) repl_broadcast gains a prev_log_term param and sets entry.term=hdr->term, entry.prev_log_index=hdr->lsn-1, entry.prev_log_term (sourced under p->lock via new entry_term_locked accessor in persist_wal_async). (2) ha_stream_wal_to_standby sets each catch-up entry's term + prev tuple (tracking the prior streamed entry's lsn/term in ascending order). (3) wal_entry_to_wire now encodes prevLogIndex/prevLogTerm/leaderCommit PER SUB-ENTRY so a multi-entry batch is library-routable. (4) ha_apply_wal_catchup_entry routes every term-carrying entry (single + each batch sub-entry) through new ha_apply_term_entry_via_library -> ha_consume_append_entries -> raft_on_append_entries_request -> on_persist_entry, which persist_set_term(leader_term) before persist_wal_write_replicated (bug-850 contiguity guard preserved beneath). The term-0 legacy direct-apply is retired for live/catch-up entries. ACK/commit path (3.3) unchanged.

## Verification
Unit RED->GREEN: test_broadcast_carries_leader_term_and_prev_tuple (repl: neuter term -> Expected 7 Was 0) + test_live_broadcast_receive_persists_leader_term (adapter: neuter receive route -> Expected 7 Was 3, i.e. standby persisted its LOCAL term). Full controller suite 175 Tests 0 Failures (clean /tmp direct binary) + meson Ok:1 Fail:0. Live STRICT_TERM gate (clean_failover_preservation.sh) NOT runnable in this overlayfs sandbox: baseline-confirmed no failover completes (split-vote no-promotion within 20s + fdatasync errno=5) — bug-924 env limitation, requires real-fs. The 7 test_wal_catchup_* (test_config_repl.c) + occasional durability FAILs on the direct binary are the SAME pre-existing bug-924 overlayfs fdatasync artifact (identical on baseline).

## References
- Related: bug-990, bug-851