Monitoring

Alerting & Notifications

An alert rule is an LPQL search that runs on a schedule. When its results meet the condition, LogPulse notifies the channels you chose.

Overview

You find alert rules under Configure → Alert Rules. A rule has a query, a time range, a schedule, a condition, a severity and one or more notification channels. You can also create a rule straight from a search or a dashboard panel, with the query already filled in.

Tip: Test the query before you save: the rule form shows what the query returns now and whether the condition would fire.

Create an alert rule

Example
Name:        API errors
Query:       index=web status>=500
Time range:  last 5 minutes
Schedule:    every minute
Condition:   number of results greater than 50
Severity:    Warning
Channels:    Slack #ops, email [email protected]

Conditions

ConditionFires whenOperators
Number of resultsThe number of rows the query returns meets the threshold>, <, =, ≠, >=, <=
Field value matchAt least one row has a field that matches the valueequals, not equals, contains, does not contain, >, >=, <, <=

With stats the query returns one row per group, so the number of results counts groups, not events. To alert on a count, compute it in the query and compare it as a field value:

LPQL
index=web status>=500 | stats count by host | where count > 100

The comparisons >, >=, < and <= on a field compare numbers and skip rows where the field is not a number.

Schedule and time range

The time range is how far back each run looks, such as the last 5 minutes. The schedule is how often it runs: an interval from every minute to daily, or a cron expression such as 0 9 * * 1-5 for weekdays at 9:00. Keep the time range at least as long as the interval, so no events fall between two runs.

Severity

SeverityWhat it does
CriticalA real problem. Notifies and shows the rule as triggered
WarningNeeds a look. Notifies and shows the rule as triggered
InformationalA report. Notifies on every hit, but never becomes a triggered alert or an incident. Use it for a daily digest

Notification channels

A rule can notify several channels at once:

ChannelWhat it does
EmailAn email to one or more addresses, with the rule, the condition and the first results
SlackA message through a Slack incoming webhook
WebhookAn HTTP POST with JSON to a URL of your choice. Use it for Microsoft Teams, PagerDuty or your own tooling
Web PushA browser notification to the members of your organization
Mobile PushA notification in the LogPulse iOS app, with a Live Activity on the lock screen until the alert resolves
Run pipelineStarts an automation pipeline, on a trigger and optionally on resolve, for example to open or close a ticket
Note: There is no built-in PagerDuty or Teams channel for alert rules. Send to them with a webhook. PagerDuty is available as a channel for anomaly detection.

Templates

Email, Slack and webhook messages can use your own template in Mustache. The variables are {{alertName}}, {{alertDescription}}, {{severity}}, {{query}}, {{timeRange}}, {{condition}}, {{conditionLabel}}, {{threshold}}, {{eventCount}}, {{triggeredAt}}, {{alertUrl}} (opens the query in Search) and {{results}}.

Webhook payload

Without a template, a webhook gets this JSON:

JSON
{
  "type": "alert_triggered",
  "alert": {
    "id": "…", "name": "API errors", "severity": "warning", "description": "…",
    "query": "index=web status>=500", "timeRange": "-5m",
    "condition": "gt", "threshold": 50
  },
  "result": { "eventCount": 73, "triggeredAt": "2026-11-08T14:02:00.000Z", "results": [ … ] }
}

A report (Informational) sends "type": "alert_report". A template that renders to JSON is sent as is; any other text arrives in a message field next to the same data.

Triggered, re-alert, resolved

MomentWhat happens
The condition is met for the first timeThe rule becomes triggered, the matching rows are kept in its history and every channel is notified
Still met at the next runNo new notification, unless you set Re-alert (every 5 minutes to every 24 hours)
No longer metThe rule is resolved by itself. Only pipeline channels set to run on resolve hear about it
The query failsThe rule shows an error with the message, and the next run tries again

Acknowledge and mute

Acknowledge an alert to say someone is on it: there are no repeat notifications until it resolves. In the iOS app you can also mute the pushes of one rule for yourself, for up to 30 days. Through the MCP server an assistant can mute a rule for the whole organization for 1 hour, 4 hours or until tomorrow morning: the rule keeps running and recording, but notifies nobody.

Access

Creating a rule needs alerts:create. A rule searches with the access of the person who created it, so it only sees the indexes that person may search.

Other kinds of alerts

KindWhere
Deviations from the normal pattern of a serviceAnomaly detection
Service health, KPIs, health checks and SLOsService Intelligence
Security detections and notablesSecurity monitoring

SLO burn rate alerts

An SLO on a service measures good against all events (two LPQL searches) or the share of KPI periods that were healthy, over a window of 7, 28, 30 or 90 days. With burn alerts on, LogPulse tells the service's notification channels when the error budget is being spent too fast, and again when the burn stops. It uses the multi-window, multi-burn-rate rules from the Google SRE workbook:

AlertBurn rateLong windowShort windowMeans
Fast burn14.4x1 hour5 minutes2% of a 30-day budget spent in one hour. Can open and resolve an incident
Slow burn6x6 hours30 minutes5% of a 30-day budget spent in six hours

Both windows have to agree, so an alert stops minutes after the problem does. A new alert waits at least 30 minutes after the previous one, maintenance windows are respected, and a window without data is never treated as an all-clear. An SLO is at risk when less than 10% of its budget is left.

API

CallWhat it does
GET /api/v1/alertsAll alert rules
POST /api/v1/alertsCreate a rule
GET / PUT / DELETE /api/v1/alerts/:idRead, change or delete a rule
POST /api/v1/alerts/:id/toggleTurn a rule on or off
POST /api/v1/alerts/test-queryRun a query and condition without saving
POST /api/v1/alerts/:id/runRun the rule at the next pass, within a minute
GET /api/v1/alerts/:id/historyWhen the rule fired and resolved
POST /api/v1/alerts/:id/acknowledgeAcknowledge the open trigger
Example
curl -X POST https://api.logpulse.io/api/v1/alerts \
  -H "Authorization: Bearer $LOGPULSE_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "API errors",
    "query": "index=web status>=500",
    "timeRange": "-5m",
    "scheduleType": "interval",
    "scheduleValue": "1m",
    "conditionType": "row_count",
    "condition": "gt",
    "threshold": 50,
    "severity": "warning",
    "notifications": [{ "type": "email", "config": { "emails": ["[email protected]"] } }]
  }'