Notification Policies
Route alerts to the right team with matching rules and escalation
What are Notification Policies?
Notification policies determine how alerts are routed and escalated. When an alert arrives, NearIRM evaluates policies in priority order until one matches. The matching policy's escalation steps then execute.
Policy Components
Matching Rules
Rules determine which alerts a policy handles. A policy matches if ALL its rules are satisfied.
| Rule Type | Description | Example |
|---|---|---|
| Severity | Alert severity level | severity = critical |
| Label | Key-value label match | env = production |
| Source | Integration source | source = grafana |
No rules? A policy with no matching rules acts as a catch-all, matching any alert.
Priority
Policies are evaluated in priority order (lowest number = highest priority). When multiple policies could match an alert, the highest priority policy wins.
Best practice: Create specific policies with low priority numbers, and a catch-all policy with a high number as a fallback.
Escalation Steps
Each policy has one or more escalation steps that execute in sequence:
- Step 1 - Notify the target immediately when alert fires
- Step 2 - If not acknowledged within timeout, notify next target
- Step 3 - Continue escalating until acknowledged or resolved
Step Targets
Each step targets one of:
| Target Type | Behavior |
|---|---|
| Team | Notifies all members of the team |
| Schedule | Notifies whoever is currently on-call |
| User | Notifies a specific user directly |
Recommendation: Use schedules for most policies. This ensures alerts reach whoever is on-call without policy updates when rotations change.
Step Timeout
Timeout defines how long to wait for acknowledgment before escalating:
- 5 minutes - Critical alerts needing immediate response
- 15 minutes - Important but not emergency
- 30 minutes - Standard response time
Creating a Policy
- Go to Policies in the sidebar
- Click Create Policy
- Enter a name and optional description
- Add matching rules (or leave empty for catch-all)
- Add escalation steps with targets and timeouts
- Set priority relative to other policies
- Save
Example: Production Critical Alerts
A policy for critical production alerts:
Matching Rules:
- Severity = critical
- Label: env = production
Escalation Steps:
- Notify "On-Call" schedule, 5 min timeout
- Notify "Engineering Leads" team, 10 min timeout
- Notify "Engineering Manager" user, no timeout
Suppression Windows
Suppress alerts during planned maintenance:
- Go to Policies
- Click Suppression Windows
- Create a window with start/end times
- Matching alerts during this window won't trigger notifications
Use suppression for:
- Scheduled maintenance windows
- Known noisy alert periods
- Deployments that trigger expected alerts
Policy Ordering
Drag policies to reorder by priority, or edit the priority number directly. Remember:
- Lower number = higher priority
- First matching policy wins
- Have a catch-all policy as backup