Every metric. One explorer.
Agent, integration, Prometheus and OpenTelemetry metrics land in one explorer. Combine queries with formulas, group by any tag and alert when a host crosses a line.
Two queries. One formula.
Divide failed requests by all requests to get an error rate, then split it by host.
- a
checkout.http_requests_totalfrom code:5xxsum - b
checkout.http_requests_totalfrom everythingsum - ƒ
a / b * 100
- a / b * 100 {everything}peak 1.48 %
The formula turns two request counts into one error rate. Sample data
Find a metric by name.
Type part of a name, keep the matches as chips and see each metric's type, hosts and tags.
- name: kafka.consumer.records_lag
- name: kafka.consumer.assigned_partitions
Click a row to drop it. Clear the chips to list every metric.
Four sources. One clock.
Point at any chart. Every chart moves to the same minute, whichever source sent it.
An average can hide a host.
Infrastructure rules run per host, so one hot server gets its own incident.
When CPU Used Percentage is above 80 % or 90 % for 15 minutes, notify the channels
db-1 gets its own incident. The other hosts stay quiet.
Sample data · 14:00 to 15:00
Questions about metrics
Pricing, deployment, data and the agent. If something is missing, a person will answer it on a demo call.
Custom metrics are metrics Atatus does not collect on its own, such as the numbers your app exposes on a Prometheus endpoint or sends through OpenTelemetry. They sit next to the built-in host metrics, so you can chart and alert on them the same way.
Scrape a Prometheus-compatible endpoint with the Atatus Infrastructure Agent, send OTLP metrics from an OpenTelemetry SDK or Collector to otel-rx.atatus.com, or turn on a bundled integration such as RabbitMQ, Cassandra or CockroachDB. Everything these collect is stored as custom metrics.
Yes. In the Prometheus integration, metrics_filters takes include and exclude name patterns such as http_* or go_*, so only the metrics you want are stored. The namespace setting adds a prefix to every metric name, which keeps metrics from different services apart.
Retention follows your plan. When the retention period ends, the data is purged from Atatus and cannot be restored, while new data keeps being collected under the same policy.
Yes. Custom metrics are one of the alert sources, next to APM, infrastructure, logs and others. A rule has an optional warning and a critical threshold and an evaluation window from 5 to 60 minutes, and notifies channels such as email, Slack, Microsoft Teams, PagerDuty or a webhook.
The Atatus Infrastructure Agent reports system metrics such as CPU, memory, swap, load average, disk and inode usage, network bytes, packets, errors and drops, and per-process CPU, memory and open file descriptors. You can chart them and alert on them per host.