Connect PagerDuty

Page your on-call team through PagerDuty whenever a monitor goes down, right from InfraNest.

Overview

• A monitor going down triggers a PagerDuty incident automatically; when it recovers, InfraNest resolves that same incident.
• Acknowledging or resolving in one tool updates the other, once you're connected.
• You only need a PagerDuty Integration Key — no REST API key or OAuth required.

Before you start

You need a PagerDuty Integration Key (sometimes called a routing key). This comes from an Events API v2 integration on one of your PagerDuty services — it is not an "API Access Key".

1. In PagerDuty, go to Services and open the service you want to be paged for (or create one).
2. Open the service's Integrations tab and choose Add integration.
3. Search for Events API v2 and add it.
4. Copy the Integration Key it shows — a 32-character key. That's the value InfraNest needs.

WarningDon't use the API Access Key from Integrations → API Access Keys (keys like PSOJL90…). That's for PagerDuty's REST API and InfraNest will reject it. Always use the Events API v2 Integration Key from a service.

Connect PagerDuty

1. Go to Alerts, then Destinations.
2. Select Add a destination and choose PagerDuty.
3. Paste the key into Integration key (routing key). InfraNest checks it when you save: a PagerDuty Integration Key is 32 letters and digits, and anything else is refused straight away rather than saved and left to fail later.
4. Select Send test to fire a test page and confirm an incident appears in PagerDuty (you can resolve it afterwards).
5. Select Save & test.

Send alerts to PagerDuty

Once the destination is saved, choose PagerDuty when you decide which alerts go where. A monitor going down triggers a PagerDuty incident (critical); when it recovers, InfraNest resolves that same incident.

Acknowledge and resolve — both ways

• Acknowledge or resolve in InfraNest → the linked PagerDuty incident is acknowledged/resolved too.
• Acknowledge or resolve in PagerDuty → the InfraNest incident updates to match.

The second direction (PagerDuty → InfraNest) is an optional one-time setup on your PagerDuty destination — look for Set up two-way sync on the destination. Paging and outbound updates work immediately with just the Integration Key.

Tips

• The Integration Key is stored securely and hidden after saving.
• One destination pages one PagerDuty service. Add another destination to page a different service or team.
• Automations can page PagerDuty too, for things a monitor never sees — an expiring certificate, a failed job.

NoteYou don't need a PagerDuty REST API key or OAuth for paging or two-way sync — only the Integration Key.

Troubleshooting

InfraNest won't accept the key. The key you pasted isn't the right shape for an Integration Key, which is always 32 letters and digits. Most often it's a REST API Access Key (they look like PSOJL90… and are much shorter) or a key with a dash in it. Follow Before you start above to get the Integration Key from your service's Integrations tab, and paste that instead.

Test failed with an "invalid routing key" message. This is the same problem from the other end — a destination saved before InfraNest started checking the key. Open the destination, paste the Events API v2 Integration Key from your service, and save.

Send test says Test sent successfully, but nothing arrives in PagerDuty. A key that is the right shape but belongs to a deleted integration is accepted by PagerDuty and then goes nowhere — PagerDuty replies "received" either way, so nothing on our side can tell. Open the service's Integrations tab in PagerDuty, confirm the Events API v2 integration is still there, and copy its key again.

Nothing pages. Check the destination is Enabled and that the monitor's alerts are routed to it. Select Send test again to confirm the key works.

A real alert didn't create a new PagerDuty incident. PagerDuty groups repeated alerts for the same issue into one open incident until it's resolved — so a flapping monitor updates the existing incident rather than paging again. That's expected; resolve the incident (or wait for recovery) and the next occurrence pages fresh.