Alerting
Alerting turns Zeotap’s operational events into real-time notifications. You define rules that watch for problems — failures, slow runs, dropped row counts, stalled pipelines — and when a rule’s condition is breached, an incident opens and notifies your channels (Slack, email, or a webhook).
Incidents recover automatically on the next clean run, and you are only notified on a state change — so you get told when something breaks and when it is fixed, without repeat noise while it stays broken.
How Alerting Works
- Define channels — where notifications go (Slack, email, or a webhook). See Channels.
- Create rules — what to watch for, on which resources, and how serious it is. See Rules.
- Incidents open on breach — when a rule’s condition is met, an incident opens and its channels are notified. See Incidents.
- Incidents recover automatically — the next clean run closes the incident and sends a recovery notification.
Alerts are driven by the same observability event stream that powers the in-warehouse log and external export — so a rule can watch any operational activity across the platform.
Where to Find It
The Alerts section in the left sidebar has three areas:
- Incidents — the current and past incidents raised by your rules
- Rules — the rules you have defined
- Channels — the destinations notifications are sent to
Key Concepts
Rules
A rule is a scoped trigger plus a condition. The scope picks which resources it watches (a resource type, specific resources, or everything); the trigger and condition define what counts as a problem — for example, three failures in a row, or a run that writes fewer rows than expected. Each rule has a severity (warning or critical) and one or more channels to notify. See Rules.
Channels
A channel is a delivery destination — a Slack webhook, one or more email recipients, or a generic webhook for tools like PagerDuty or Opsgenie. See Channels.
Incidents
An incident is a single open problem for a rule and resource. Repeated breaches while an incident is open increment its occurrence count rather than sending a fresh notification. Incidents can be muted for an hour, a day, or indefinitely to suppress notifications during known work. See Incidents.
Next Steps
- Rules — Define what to watch for
- Channels — Set up Slack, email, and webhook delivery
- Incidents — Manage open incidents and muting
- Observability — The event stream behind alerting