On-call

Wake the right engineer. Let everyone else sleep.

The agent responds first. Schedules and escalations are the last line of defense — and they don't stop at voicemail.

Rules you can read aloud

The policy is the sentence. The ladder is yours. No YAML archaeology at 3am.

Escalations in plain English

If nobody acknowledges in five minutes, call the on-call. Then the next person up. Then the team admins — with repeat loops and auto-close when every step has run.

You decide how you get woken

Call after one minute, SMS after three, email after five. Quiet hours hold low-priority noise until morning — each responder sets their own ladder.

The life of a page

From alert to acknowledged — what happens when it's real.

1

The alert gets routed

Severity, tags, source, message — routing rules send each alert to a schedule or straight into an escalation policy.

2

The phone rings — and rings on

Warrn calls, listens for a human, and takes acknowledge or resolve from the keypad. An answering machine is not an answer — the escalation moves to the next person.

3

A human takes it from here

Acknowledged, escalation stopped, timeline updated — and the investigation the agent started is already waiting in the alert.

What you open when you answer

Acknowledge from the call, open the alert — the agent's investigation, suggested actions, and the alert's full context are already there. On your laptop or your phone.

Investigation Complete

6/24/2026, 2:58:06 PM

95%Request Re-run

The healthy-host-count alarm fired because a Terraform-managed rolling deployment inadvertently kept HealthyHostCount at 1 for 5 consecutive minutes (13:55–13:59 UTC), tripping the 4-minute sustained threshold designed to distinguish real outages from normal rolling-deploy churn. CloudTrail shows Terraform deregistered warrn-api-1 at 13:55:12, re-registered it at 13:58:00, then deregistered warrn-api-2 only 21 seconds later — before api-1 had passed its ALB health checks — so the count never recovered to 2 between the two operations. CloudWatch fired ALARM at 14:01:22 and auto-resolved to OK at 14:03:22. Both EC2 instances remained running with passing status checks throughout; no crash, no stop-instance event, and no 5xx errors were observed. The service was degraded to a single host for ~5 minutes but was never completely down.

ACTIONS0/4 done

Increase inter-step delay in Terraform rolling deploy

Soon

Insert a minimum wait of 90–120 seconds between RegisterTargets for instance N and DeregisterTargets for instance N+1, keeping HealthyHostCount at 2 throughout the deploy.

Add a health-check gate before proceeding to next deploy step

When Possible

Have the deploy script poll DescribeTargetHealth after each RegisterTargets and only proceed once HealthyHostCount returns to 2.

Review alarm evaluation_periods vs. worst-case deploy overlap

When Possible

Consider raising evaluation_periods to 6 to add headroom, or suppress the healthy-hosts alarm during active Terraform runs.

The agent does the repetitive work. You make the calls.

On-call as a collaboration — with guardrails that keep anything touching production behind a human.

A fixed first response, per alert type

Encode the steps once. Every alert of that type gets the same investigation before a human is paged — and anything that writes waits for approval.

Wake up to a hypothesis, not a wall of graphs

One or more suspected causes, each with the evidence behind it. You check the reasoning, not the dashboards.

Work with it like a colleague

On the phone, on the web, in Slack, or in the warroom on a Meet. Same agent, same context, wherever you already are.

Rules you talk into existence

Describe the routing you want and the alert engine writes the if-this-then-that for you. No more juggling conditions by hand.

COMING SOONwarrn cli·MCP server·Terraform— so on-call response fits how you work.

The boring parts, done properly

Overrides & covers

Swap a shift in one click — cover a single rotation or the whole schedule.

In your calendar

Every schedule is an iCal feed. Shifts sit next to your meetings.

Shift reminders

A nudge before every shift starts — push and email, on your schedule.

Layers, restrictions & honest history

Stack a business-hours rotation over a 24/7 one. Handoffs stay correct across DST — and editing a rotation never rewrites who was actually on call.

“Who's on call?”

Answered in the sidebar with your shift countdown — and in Slack.

Start with your noisiest service

Point one alert source at Warrn and watch it get quiet.

  • 14-day free trial
  • No credit card required

Starting at

$0/month

Free for small teams up to 5 users

Start Free Trial