The record of each incident, and the first draft of its postmortem.
SKYLINE keeps a light record of each incident: what happened, who handled it and how long it took. It is not a paging tool; it gives you the timeline and the numbers.
Opened by alerts, or by hand
An alert group opens an incident. When it fires again it joins the open one, so a storm is one incident.
Four states, one owner
Triage, active, mitigated, resolved. Alertmanager can only mitigate; a person resolves.
Checked after mitigation
5, 15 and 30 minutes later SKYLINE reads the namespace again. If something is still broken, the incident reopens.
MTTA, MTTM, MTTR
Medians over the last 30 days, each counted from detection.
From an incident to its postmortem
- 01
Investigate in chat
A chat opens with the environment picked. Add its diagnosis to the incident in one click.
- 02
Agree or dispute
Members mark each investigation. A root cause more people dispute than agree with is shown as disputed.
- 03
Write the draft
A blameless postmortem written from the record only, in the language of your console. Send it to GitLab or Jira as an issue.