Skip to main content

Monitor and debug apps

You’ll find why an app misbehaves from the app detail page — status, instances, logs, and terminal — without a local kubectl workflow.

Check first​

  1. Status / health of the app.
  2. Copies vs instances — if you asked for 2 copies but only 1 instance runs, something is failing to start.
  3. Recent changes (image tag, variables, resource limits).

Logs​

  1. Open the app → Logs.
  2. Select an instance if more than one is running.
  3. Watch live output while you reproduce the problem.

Empty stream usually means the process is not writing to stdout/stderr, or it crashes before printing.

Terminal​

Open Terminal for a shell inside a running instance. Prefer read-only checks on production unless you know the impact.

Restart and pause​

  • Restart recreates running instances (useful after a stuck state).
  • Pause stops the app without deleting it (useful overnight for non-prod).

Deeper signals​

  • Monitoring for CPU and memory
  • Argus when the public URL is down
  • AI agent for guided troubleshooting
  • Logging when you need a dedicated log walkthrough