Infrastructure monitoring

Infrastructure alerts that respect severity

Hook your infrastructure monitoring (Prometheus, Zabbix, Uptime Kuma, custom) into WardenPoint. Severity decides channel; labels decide responder; resolve hooks cancel the chain.

Start free Browse integrations

Multi-source intake Label-aware routing Resolve hook cancels

What a quiet day looks like

10:00
Prometheus
Warning: node memory 78% on app-03
10:00
WardenPoint
Telegram message to ops channel · no escalation
14:21
Prometheus
Resolved · memory back to baseline
14:21
WardenPoint
Audit log marks 'resolved' · no further dispatches

Where infra teams hurt

Infra monitoring pain

Three complaints from infrastructure teams whose alerts wake people for the wrong reasons.

Every alert pages on-call

Without severity-aware routing, warning-level alerts wake people up for non-incidents.

Fix

Map severity to channel: warnings go to chat, criticals go to voice. Severity is the only knob.

Flapping monitors create storms

A flaky node fires alert/resolve/alert/resolve until someone silences it manually.

Fix

We dedupe by source-and-key. Identical alerts within a window collapse into one notification with a counter.

No way to mute by host group

When a cluster is in maintenance, alerts from that cluster should not page anyone.

Fix

Maintenance windows mute by label match. Set a TTL and we resume normal routing automatically.

Escalation policy

Recipe: severity-aware routing

Three severity tiers, three different paths. Configure once, reuse everywhere.

1
At minute
0:00
Critical → primary on-call by voice
Severity=critical fires phone + Telegram voice immediately. Cluster down, payments broken, customer-impacting outage.
voice_calltelegram_voice
2
At minute
0:00
Warning → ops chat
Severity=warning posts to the ops Telegram chat. No one wakes up; the next-shift handles it.
telegram
3
At minute
0:05
Critical unacked → secondary
If five minutes pass without ack, secondary on-call gets the phone call.
voice_call
4
At minute
0:00
Info → email digest
Info-level alerts collapse into a daily email digest. Useful for trends, not for action.
email

Where alerts come from

Common infra sources

Whatever you run, WardenPoint accepts the webhook.

Prometheus

Alertmanager receiver

What changes

What infra teams report

Sleep restored for warnings

Warning-level alerts move out of the on-call phone and into the chat. Sleep ratings improve.

Dedup tames flap storms

Identical alerts collapse. Post-mortems no longer count 30 pages for one incident.

Maintenance is honest

Maintenance windows mute by label. No one apologises for paging during planned downtime.

Infra FAQ

Infrastructure-team questions

How do you dedupe identical alerts?: We hash (source, alertname, labels) and treat repeats within a configurable window as the same incident. The audit log keeps a counter; only the first one fires the chain.
Can I silence a label range?
Does WardenPoint support inhibition rules?
Can I see flap rate per source?
What about alert grouping by team?

Free plan

Make infra alerts respect severity

Wire one Prometheus or Zabbix source, define severity routing and watch the next warning land in chat — not in the bedroom.