Skip to content
Install

Metrics and alerts

Orchard tells you how your apps and databases are doing in three ways. Metrics show how a resource behaved over time, the list of current problems shows what is wrong now, and alert rules notify you when you are not looking at the dashboard.

Each app shows CPU and memory as a current value and over time on its Observe tab. Databases add size, connections, and activity. The panels on each tab are listed in Monitor an app and Monitor a database.

When you size an app, read the range. kubectl top reports one instant, which can mislead you on a workload with spikes. To size from recent use instead, apply resource suggestions.

A project’s Overview tab, and the top of each app and database page, show a list of what is wrong now, with the worst problem first. The list is hidden when nothing is wrong.

Problem Applies to
Killed for using too much memory Apps and databases
Crash loop Apps and databases
Failed build Apps
Rolled-back release Apps
A change the cluster refused Apps
Firing alert Apps
Error spike in the logs Apps
Fewer replicas ready than requested Apps

Each row names the resource, shows how long the problem has existed, and links to the action that helps.

Alerts above an app’s tabs reflect the current state of the cluster. They include a failing build, a crash-looping pod, resource pressure, and a lost live connection. An alert clears when its condition clears. Short grace windows apply where state can change quickly, so a single restart does not leave a permanent banner.

To receive notifications when you are not viewing the dashboard, use these features:

Organization-wide usage, by project and by person, is on the Usage page.