Skip to content

alert: silence non-Mobula head_lag stale alerts - #922

Merged
Flotapponnier merged 1 commit into
devfrom
fix/mobula-only-alerts
Jul 6, 2026
Merged

alert: silence non-Mobula head_lag stale alerts#922
Flotapponnier merged 1 commit into
devfrom
fix/mobula-only-alerts

Conversation

@Flotapponnier

Copy link
Copy Markdown
Collaborator

Feedback

'on veut dans l'alerting seulement si ca concerne mobula; le reste on s'en fou' — 2026-07-06

Change

Rename AggregatorHeadLagStaleMobulaHeadLagStale and filter on aggregator="mobula". Non-Mobula aggregators (gecko, codex) are tracked via log tailing only.

Rationale

The old rule fired every few hours on gecko/bnb (dead WBNB/BUSD pool — separate fix in #920) and periodically on codex (WS disconnects). Both are third-party issues Mobula operators cannot act on.

Kept

  • alertmanager routing labels (severity, alert_type, app) unchanged
  • MissingHeadLagMetrics (global monitor-down) kept as-is
  • CodexServiceDown kept (misleading name but it monitors the whole latency_monitor service via up metric, not codex specifically)
  • All Mobula* alerts unchanged

Rename AggregatorHeadLagStale to MobulaHeadLagStale and filter the
expression on aggregator='mobula'. Non-Mobula aggregator health
(gecko, codex, ...) is tracked via log tailing, not paged. Operator
feedback 2026-07-06: 'on veut dans l'alerting seulement si ca concerne
mobula; le reste on sen fou'.

The rule previously fired every few hours on gecko/bnb because the
WBNB/BUSD pool went idle in 2024 (separate fix #920 replaces that
pool with WBNB/USDT V3), and periodically on codex due to WS
disconnects — both being third-party issues Mobula operators cannot
act on.

Prometheus alert_type label + severity stay identical so the
alertmanager routing / silencer rules keep working.
@Flotapponnier
Flotapponnier merged commit a56e2f4 into dev Jul 6, 2026
1 check passed
Flotapponnier added a commit that referenced this pull request Jul 6, 2026
Rename AggregatorHeadLagStale to MobulaHeadLagStale and filter the
expression on aggregator='mobula'. Non-Mobula aggregator health
(gecko, codex, ...) is tracked via log tailing, not paged. Operator
feedback 2026-07-06: 'on veut dans l'alerting seulement si ca concerne
mobula; le reste on sen fou'.

The rule previously fired every few hours on gecko/bnb because the
WBNB/BUSD pool went idle in 2024 (separate fix #920 replaces that
pool with WBNB/USDT V3), and periodically on codex due to WS
disconnects — both being third-party issues Mobula operators cannot
act on.

Prometheus alert_type label + severity stay identical so the
alertmanager routing / silencer rules keep working.

Co-authored-by: Florent Tapponnier <contact@mobula.io>
@Flotapponnier
Flotapponnier deleted the fix/mobula-only-alerts branch July 17, 2026 14:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant