Skip to main content
When something is wrong, ask Monk to look. It reads workload status and logs from your cluster, and your coding agent can then fix the code or the MANIFEST.

​
Before you start

​
Ask Monk

Status:
Logs:
Monk returns a bounded tail of the logs (up to 2,000 lines) and can follow new lines for a short time. Logs can contain sensitive output, so Monk quotes only the lines that matter. Describe the problem:
Monk checks status, logs and the running configuration, and tells you what it found. If the cause is in your code, your coding agent makes the fix. Monk does not edit application code. Template problems:

​
What you review

Reading status and logs needs no approval. Any fix that changes infrastructure, such as a redeploy, a restart, a config change or a new node, opens an approval in the local dashboard first.

​
Redeploy or stop a service

Redeploying, stopping, deleting or purging a workload opens an approval. Purging also removes its data.

​
Limits

  • Your coding agent can’t open a shell in a container. You can, from a terminal: monk shell <workload> opens one, and monk exec runs a single command.
  • For CPU, memory and disk use per workload or per node, run monk stats or monk cluster stats in a terminal. Monk doesn’t collect application metrics such as queue depth or slow queries. For alerts on crashes and resource use, set up Watcher.

​
Still stuck?

This sends a report with environment details to the Monk team. There is no reply channel.

​
Next