actualbudget+monitoring: per-account bank-sync metrics, drop noisy alert

The bank-sync CronJob was posting to /accounts/banksync which fans out to
ALL accounts in a single call. With PSD2/GoCardless's 4-successful-pulls
per-account per-24h quota, a single rate-limited account would 500 the
whole call, and `bank_sync_success` would flip to 0 even though the data
itself was still flowing through manual UI syncs. Result: BankSyncFailing
fired routinely whenever the user had been active in the UI that day —
a structural false positive.

Fix:
  * CronJob: enumerate accounts via GET /accounts, POST per-account
    /accounts/{id}/banksync, emit bank_sync_account_success and
    bank_sync_account_last_success_timestamp labelled by account name.
    Roll up bank_sync_success = 1 iff any account succeeded.
  * Alerts: drop BankSyncFailing (noise generator). Keep BankSyncStale
    at 48h (global drought). Add BankSyncAccountStale at 72h (catches
    single-account auth expiry — the real signal we wanted).

Verified: manual run on bank-sync-viktor pushes 6 per-account success +
timestamp series; roll-up bank_sync_success=1; no firing alerts.

stacks/actualbudget/factory/main.tf

@@ -282,48 +282,93 @@ resource "kubernetes_cron_job_v1" "bank-sync" {
           spec {
             container {
               name  = "bank-sync"
-              image = "curlimages/curl"
+              image = "alpine:3.20"
               command = ["/bin/sh", "-c", <<-EOT
-              PUSHGATEWAY="http://prometheus-prometheus-pushgateway.monitoring:9091/metrics/job/bank-sync-${var.name}"
+              set -u
+              apk add --no-cache curl jq >/dev/null 2>&1
+
+              USER_NAME='${var.name}'
+              SYNC_ID='${var.sync_id}'
+              API_KEY='${random_string.api-key.result}'
+              PW='${var.budget_encryption_password}'
+              PG="http://prometheus-prometheus-pushgateway.monitoring:9091/metrics/job/bank-sync-$USER_NAME"
+              API="http://budget-http-api-$USER_NAME"
+
               START=$(date +%s)
 
-              HTTP_CODE=$(curl -s -o /tmp/response.txt -w '%%{http_code}' \
+              # Enumerate active accounts: open + on-budget.
-                -X POST --location \
+              ACCOUNTS=$(curl -fsS "$API/v1/budgets/$SYNC_ID/accounts" \
-                'http://budget-http-api-${var.name}/v1/budgets/${var.sync_id}/accounts/banksync' \
+                -H "x-api-key: $API_KEY" \
-                --header 'accept: application/json' \
+                -H "budget-encryption-password: $PW" \
-                --header 'budget-encryption-password: ${var.budget_encryption_password}' \
+                | jq -c '.data[] | select(.closed == false and .offbudget == false) | {id, name}')
-                --header 'x-api-key: ${random_string.api-key.result}')
 
-              END=$(date +%s)
+              if [ -z "$ACCOUNTS" ]; then
-              DURATION=$((END - START))
+                echo "ERROR: GET /accounts returned no eligible accounts; aborting"
-
+                exit 1
-              if [ "$HTTP_CODE" = "200" ]; then
-                SUCCESS=1
-                LAST_SUCCESS=$END
-              else
-                SUCCESS=0
-                echo "Bank sync failed with HTTP $HTTP_CODE:"
-                cat /tmp/response.txt
-                echo ""
               fi
 
-              # Pushgateway POST preserves metrics not in the payload, so on
+              : > /tmp/payload
-              # failure we omit bank_sync_last_success_timestamp to keep the
+              rm -f /tmp/any_success
-              # prior success value — this prevents BankSyncStale from firing
+
-              # alongside BankSyncFailing after a single failed run.
+              # Per-account sync. Each account has its own PSD2/GoCardless
-              {
+              # quota (4 successful pulls per 24h), so we treat them
-                printf '# HELP bank_sync_success Whether the last bank sync succeeded (1=ok, 0=fail)\n'
+              # independently — one rate-limited account doesn't mark the
-                printf '# TYPE bank_sync_success gauge\n'
+              # run as a failure.
-                printf 'bank_sync_success %s\n' "$SUCCESS"
+              echo "$ACCOUNTS" | while IFS= read -r ACCT; do
-                printf '# HELP bank_sync_duration_seconds Duration of the last bank sync run\n'
+                [ -z "$ACCT" ] && continue
-                printf '# TYPE bank_sync_duration_seconds gauge\n'
+                ID=$(echo "$ACCT" | jq -r '.id')
-                printf 'bank_sync_duration_seconds %s\n' "$DURATION"
+                NAME=$(echo "$ACCT" | jq -r '.name')
-                if [ "$SUCCESS" = "1" ]; then
+                LABEL=$(echo "$NAME" | sed -E 's/[^a-zA-Z0-9]+/_/g')
-                  printf '# HELP bank_sync_last_success_timestamp Unix timestamp of the last successful sync\n'
+
-                  printf '# TYPE bank_sync_last_success_timestamp gauge\n'
+                HTTP_CODE=$(curl -s -o /tmp/r.txt -w '%%{http_code}' \
-                  printf 'bank_sync_last_success_timestamp %s\n' "$LAST_SUCCESS"
+                  -X POST "$API/v1/budgets/$SYNC_ID/accounts/$ID/banksync" \
+                  -H 'accept: application/json' \
+                  -H "x-api-key: $API_KEY" \
+                  -H "budget-encryption-password: $PW") || HTTP_CODE=0
+
+                NOW=$(date +%s)
+                if [ "$HTTP_CODE" = "200" ]; then
+                  echo "OK account=$NAME"
+                  printf 'bank_sync_account_success{account="%s"} 1\n' "$LABEL" >> /tmp/payload
+                  printf 'bank_sync_account_last_success_timestamp{account="%s"} %s\n' "$LABEL" "$NOW" >> /tmp/payload
+                  : > /tmp/any_success
+                else
+                  echo "FAIL account=$NAME http=$HTTP_CODE body=$(cat /tmp/r.txt)"
+                  printf 'bank_sync_account_success{account="%s"} 0\n' "$LABEL" >> /tmp/payload
                 fi
-              } | curl -s --data-binary @- "$PUSHGATEWAY"
+              done
+
+              END=$(date +%s)
+              DUR=$((END - START))
+
+              if [ -f /tmp/any_success ]; then
+                ANY=1
+              else
+                ANY=0
+              fi
+
+              # Pushgateway POST preserves prior values for label sets not
+              # in the payload, so per-account last_success_timestamp values
+              # for accounts that failed this run keep their prior good
+              # values — that's what BankSyncAccountStale alerts on.
+              {
+                printf '# HELP bank_sync_account_success Per-account sync result (1=ok, 0=fail)\n'
+                printf '# TYPE bank_sync_account_success gauge\n'
+                printf '# HELP bank_sync_account_last_success_timestamp Per-account Unix timestamp of last successful sync\n'
+                printf '# TYPE bank_sync_account_last_success_timestamp gauge\n'
+                cat /tmp/payload
+                printf '# HELP bank_sync_success 1 if at least one account synced this run\n'
+                printf '# TYPE bank_sync_success gauge\n'
+                printf 'bank_sync_success %s\n' "$ANY"
+                printf '# HELP bank_sync_duration_seconds Total duration of the cron run\n'
+                printf '# TYPE bank_sync_duration_seconds gauge\n'
+                printf 'bank_sync_duration_seconds %s\n' "$DUR"
+                if [ "$ANY" = "1" ]; then
+                  printf '# HELP bank_sync_last_success_timestamp Unix timestamp of the most recent successful sync of any account\n'
+                  printf '# TYPE bank_sync_last_success_timestamp gauge\n'
+                  printf 'bank_sync_last_success_timestamp %s\n' "$END"
+                fi
+              } | curl -fsS --data-binary @- "$PG"
               EOT
               ]
             }

stacks/monitoring/modules/monitoring/prometheus_chart_values.tpl

@@ -2152,20 +2152,24 @@ serverFiles:
               severity: warning
             annotations:
               summary: "Mail server has no available replicas - mail may not be received"
-          - alert: BankSyncFailing
+          # Note: no BankSyncFailing alert — GoCardless enforces per-account
-            expr: bank_sync_success == 0
+          # PSD2 quotas (4 successful pulls per account per 24h). Manual UI
-            for: 6h
+          # syncs consume the same quota, so the nightly cron routinely hits
-            labels:
+          # rate-limits without any real outage. Alert only on staleness.
-              severity: warning
-            annotations:
-              summary: "Bank sync failing. Accounts may need GoCardless reauthorization. Check Pushgateway for which instance."
           - alert: BankSyncStale
             expr: (time() - bank_sync_last_success_timestamp) > 172800
             for: 1h
             labels:
               severity: warning
             annotations:
-              summary: "Bank sync has not succeeded in more than 48h. Check CronJob and account auth."
+              summary: "Bank sync (instance {{ $labels.instance }}): NO account has synced in over 48h. Likely a real outage — check CronJob, http-api logs, and GoCardless re-auth."
+          - alert: BankSyncAccountStale
+            expr: (time() - bank_sync_account_last_success_timestamp) > 259200
+            for: 1h
+            labels:
+              severity: warning
+            annotations:
+              summary: "Bank sync (instance {{ $labels.instance }}): account {{ $labels.account }} has not synced in over 72h. GoCardless requisition may have expired — re-link in Settings → Bank Sync."
           - alert: EmailRoundtripFailing
             expr: email_roundtrip_success{job="email-roundtrip-monitor"} == 0
             for: 60m