Point an Alertmanager webhook receiver at OpsPing and every firing pages on-call: alertname becomes the message, severity maps to P1–P5, and the fingerprint becomes a stable alias — with send_resolved: true, resolutions close the alert automatically.
Receiver configuration
route:
receiver: opsping
group_by: [alertname]
receivers:
- name: opsping
webhook_configs:
- url: 'https://api.opsping.com/v2/webhooks/inbound/<id>/<token>'
send_resolved: trueUse an OpsPing API Key integration — it auto-detects Alertmanager v4 payloads. The standard alerts array is processed entry by entry; the webhook response tallies outcomes (created, deduped, closed).
Field mapping
| Alertmanager field | OpsPing alert |
|---|---|
labels.alertname | Message |
labels.severity | Priority — critical/error/warning/info/debug → P1–P5; missing or unknown → P3 |
fingerprint | Alias — stable across firings |
endsAt in the past / resolved entry | Closes the matching alert |
| Labels and annotations | Alert details |
One receiver per service
Alertmanager's payload has no OpsPing serviceId field, so alerts inherit the integration's default service. If one Alertmanager routes alerts for many systems, create one OpsPing integration per service and route each to its matching receiver — the alert queue and mobile app then show the right service icon for every page.
Prove the path end to end
The integration guide ships a reproducible smoke: run a real Alertmanager container, submit a critical database alert to its v2 API, and watch Alertmanager deliver the v4 webhook — OpsPing shows one open P1 sourced from your integration within seconds. Run it before cutover; it exercises grouping, delivery, and service linking, not a fixture POST.
- Noise control: match on
source, tags, or priority in alert policies to suppress or delay low-severity Alertmanager noise. - Alerts covered by a maintenance window are stored with notifications suppressed — they are not deferred-paged when the window ends.
FAQ
Which Alertmanager versions work?
The mapper targets v4 webhook payloads — the format Alertmanager has emitted since the v0.11 era, so any current Alertmanager works. OpsPing auto-detects the shape; no type flag needed.
How do resolved alerts close in OpsPing?
Set send_resolved: true on the webhook receiver. A resolved entry closes whichever alert currently owns that fingerprint's alias — even an already-acknowledged one. Without it, alerts stay open until manually closed or auto-closed by a policy.
What priority do alerts get without a severity label?
P3. Severity maps critical→P1, error→P2, warning→P3, info→P4, debug→P5; missing or unknown severity falls back to P3. Map severities in Alertmanager (or re-rank in OpsPing with alert policies) if that default is wrong for you.