Feature WalkthroughsAlerts and incident response

20. Alerts and incident response

The Alerts page is the operational inbox. Chapter 11 covered the model — events, alerts, incidents, notifications. This chapter is the UI walkthrough.

URL: https://ops.daalu.io/alerts.


The list view

The default Alerts page shows currently-firing alerts at the top, then recently-resolved ones below. Columns:

  • Severity badge — info / warning / critical.
  • Title — the alert’s headline (from the rule).
  • Source — the integration that produced it.
  • Labels — pills showing the alert’s labels (cluster, device, service, team).
  • First seen — when the alert was first noticed.
  • Assignee — who’s looking at it. Empty means nobody.
  • Status — firing / resolved / snoozed.

Top of page filters: severity (multi-select), source, status, assignee, label search, full-text search.

Top-right actions: Mute all warnings (snooze every warning for 1 hour, useful during maintenance windows) and Acknowledge all (set yourself as the assignee on every unassigned critical).


Alert detail page

Click an alert to drill in. The detail page has:

  • Title, severity, status pill.
  • Buttons: Acknowledge (assign to me), Snooze, Resolve manually, Promote to incident.

Auto-generated explanation

A paragraph or two written by the alert-explainer agent (see Chapter 19). Inputs cited at the bottom — click any to drill in.

If the explanation looks wrong, click “Send feedback.” Real human eyes on the engineering side review these and the agent’s prompts get tuned over time.

Suggested actions

A list of buttons the agent recommends:

  • “Restart deployment X”
  • “Page on-call Y”
  • “Open change proposal Z”
  • “Snooze for 1 hour while we wait for the planned maintenance”

Clicking any of these is the trigger; mutations still go through a proposal.

Timeline

The history of this alert:

  • Fired at T0.
  • Acknowledged by Jane at T+2m.
  • Comment by Bob at T+5m.
  • Resolved at T+12m.

Includes any Assistant interactions on this alert.

Comments

A threaded discussion. Each comment notifies subscribers.

Labels

The alert’s labels in a table, editable inline. Editing labels on a single alert is occasionally useful for routing (“oh, this should have been tagged team: netops”). Editing the rule that produces the alert is usually what you want; that’s in Settings → Alert rules.

A panel showing other recent alerts with overlapping labels. Click any to jump.


Promote to incident

Some alerts deserve a full incident — multiple alerts firing together, customer impact, on-call coordination needed.

Click Promote to incident on an alert to create the incident. The dialog asks for:

  • Title.
  • Severity (defaults to the alert’s).
  • Assignee (defaults to you).
  • Additional alerts to attach.

The incident gets its own page (URL /incidents/<id>) with a timeline that all attached alerts contribute to.


Incident page

The incident page extends the alert detail page with:

  • Timeline — every event (alerts joining, comments, Assistant suggestions, manual actions) chronologically.
  • Attached alerts — currently in scope of the incident.
  • Postmortem draft — auto-generated outline when the incident is resolved. You edit it and “Save as final” to preserve.

Incidents are notably less common than alerts. Most teams open 1–3 per month. They’re the unit you bring to a retrospective.


Snoozing

From any alert detail page:

Snooze for → 30 min / 2 hours / 8 hours / Custom.

What snoozing does:

  • Suppresses notifications for that alert for the duration.
  • Keeps the alert status firing (snoozing isn’t resolving).
  • Logs the snooze in the timeline with the snoozer’s name.
  • Only affects your notifications, not your teammates’.

What snoozing doesn’t do:

  • Resolve the alert. The underlying condition is still there.
  • Affect downstream PagerDuty escalation paths. (Snooze in Daalu is independent of PagerDuty’s mute, by design — you almost never want to silence both at once.)

A snoozed alert shows a “Snoozed by X until Y” pill at the top of its detail page.


Triggering a test alert

For verifying the alerting pipeline:

Settings → Alerts → Send test alert.

The dialog lets you pick severity, label set, and target channels. Daalu fires the alert end-to-end through the same pipeline a real one would use, then auto-resolves it after 60 seconds.

Useful when:

  • You just configured Slack / PagerDuty and want to confirm the routing works.
  • You’re testing a new alert rule’s labels.
  • You’re onboarding a new on-call and want them to see what a real page looks like.

Alert rules

Hover over the alert title — you’ll see the rule that produced it. Click for the rule detail page:

  • Conditions — the predicate that matches events.
  • Severity — the rule’s severity (overridable per fire).
  • Labels — labels emitted on each fire.
  • Runbook URL — link shown on every firing instance.
  • Recent fires — history of this rule firing.

Editing rules is in Settings → Alert rules and is admin-only.


Bulk operations on alerts

For when you have a storm:

  • Multi-select with checkboxes.
  • Acknowledge selected — assign yourself to all selected.
  • Snooze selected for 2 hours — bulk snooze.
  • Resolve selected — only available if you’re allowed to resolve. Use rarely; alerts are supposed to auto-resolve when the condition clears.
  • Promote selected to incident — bundle into one incident.

What’s next

Chapter 21 covers Workflows / Automations — the page where you author your own agents and scheduled jobs.