Team Playbook

Appearance

Incidents & Postmortems ​

Every incident is a Jira Task labelled incident, and its write-up and postmortem live in that Task's description. Slack is for real-time coordination; the Jira Task is the permanent record. The work an incident produces is tracked in Jira like any other work, linked to the incident Task.

Declaring an incident ​

Anyone can declare. Lower the bar — over-declaring is cheap, under-declaring is expensive.

  1. Post in #incidents: SEV<N>: <one-line summary>.
  2. Open an incident channel: #inc-YYYYMMDD-<short-name>.
  3. Assign an Incident Commander (IC) — usually the on-call. The IC coordinates; they don't fix.
  4. Update #incidents every 30 minutes (SEV1) or 60 minutes (SEV2) until resolved.
  5. Create a Jira Task labelled incident immediately, with the postmortem template as its description, and link it in the incident channel. Start filling it in during the incident — don't wait until it's resolved.

During the incident ​

  • IC drives. One person types in the channel; others execute.
  • Keep a timeline as you go. Copy-paste timestamps directly into the incident Task.
  • If you're stuck for >30 minutes on SEV1, escalate.

Postmortem ​

  • Required for all SEV1 and SEV2. Optional but encouraged for SEV3.
  • Drafted in the incident Task's description within 5 business days of resolution.
  • Blameless. Focus on the system: "Why did the system make this easy to get wrong?"
  • Reviewed with the team; the IC then comments "Postmortem reviewed" and closes the incident Task with that as its reason.
  • Action items: maximum 5, each with an owner and due date. Each is filed as a Jira Task labelled incident-followup and linked to the incident Task.