Merge pull request 'fix(monitoring): don't false-fire the WAL-stalled alert on an idle database' (#241) from fix/pg-archive-stalled-idle-falsepositive into development

deploy/README.md

@@ -255,8 +255,10 @@ shipping or redeploying this stack does **not** start archiving — the artifact
 armed, which is why an un-armed prod deploy can never pile WAL onto the disk. The base-backup
 timer is provisioned by the Ansible `main` role behind `pitr_enabled` (also default off). Two
 Grafana alerts watch health: `WAL archiving failing` (`pg_stat_archiver_failed_count` rising)
-and `WAL archiving stalled` (`pg_stat_archiver_last_archive_age` over 30 min) — both
+and `WAL archiving stalled` (last archive over 30 min old **while `pg_wal_size_bytes` is growing**
-absent/NaN-safe, so they stay quiet until archiving is armed.
+— the pg_wal-growth guard keeps an idle database, which archives nothing because it writes nothing,
+from false-triggering during quiet hours) — both absent/NaN-safe, so they stay quiet until
+archiving is armed.
 
 **Assessment (owner-reviewed; the gate before the first real money).** Measured on prod
 `pg_stat_wal`: WAL is generated at **~0.77 MB/day** and the database is **~9.6 MB**. At

deploy/grafana/provisioning/alerting/rules.yaml

@@ -245,8 +245,14 @@ groups:
               conditions:
                 - evaluator: { type: gt, params: [0] }
         labels: { severity: critical }
-        annotations: { summary: 'pgBackRest archive_command is failing — PITR is degrading and pg_wal can fill the disk. Run `docker exec scrabble-postgres pgbackrest --stanza=scrabble check`.' }
+        annotations: { summary: 'pgBackRest archive_command is failing — PITR is degrading and pg_wal can fill the disk. Run `docker exec scrabble-postgres pgbackrest --stanza=scrabble --pg1-user=scrabble check`.' }
 
+      # Fires only when the last archive is >30 min old AND pg_wal is actually growing (WAL is being
+      # produced but not archived). A genuinely idle database archives nothing — Postgres does not
+      # force-switch an empty segment on archive_timeout — so `last_archive_age` alone false-triggers
+      # during quiet hours; the pg_wal-growth guard removes that. `and on()` bridges the differing
+      # label sets (last_archive_age has a server label, the pg_wal gauge does not); delta() (not
+      # increase) suits the pg_wal_size_bytes gauge. Real archive failures are caught by pg_archive_failing.
       - uid: pg_archive_stalled
         title: WAL archiving stalled
         condition: C
@@ -255,11 +261,11 @@ groups:
         execErrState: OK
         data:
           - refId: A
-            relativeTimeRange: { from: 300, to: 0 }
+            relativeTimeRange: { from: 2100, to: 0 }
             datasourceUid: prometheus
             model:
               refId: A
-              expr: pg_stat_archiver_last_archive_age
+              expr: (pg_stat_archiver_last_archive_age > 1800) and on() (delta(pg_wal_size_bytes[35m]) > 16777216)
               instant: true
           - refId: C
             datasourceUid: __expr__
@@ -270,4 +276,4 @@ groups:
               conditions:
                 - evaluator: { type: gt, params: [1800] }
         labels: { severity: critical }
-        annotations: { summary: 'No WAL segment archived for over 30 minutes (archive_timeout is 5m) — archiving is stalled; the PITR recovery point is frozen and pg_wal may grow.' }
+        annotations: { summary: 'No WAL archived for over 30 minutes while pg_wal keeps growing — archiving is genuinely stalled; the PITR recovery point is frozen and pg_wal will fill the disk. Run `docker exec scrabble-postgres pgbackrest --stanza=scrabble --pg1-user=scrabble check`.' }