Troubleshooting

Coming in the first release. This page describes behaviour being built for the first Gliss release. Details may change before it ships.

Every diagnosis names a cause, whose problem it is, and who should act next.

Whose problem it is

Responsibility	Meaning	Who acts
`app`	The app’s code or `gliss.yaml`	The app owner, usually through their AI
`platform`	The cluster or a Gliss component	A maintainer or IT. The builder’s AI cannot fix it.
`unknown`	Gliss could not tell	Start with the summary and the logs

Cause	Typical meaning
`crash-loop`	The app starts, then exits, again and again
`oom-killed`	The app ran out of memory
`image-pull`	The cluster cannot pull the app’s image
`probe-failure`	The app runs but does not answer health checks
`pending`	The cluster cannot place the app; often capacity. See Capacity.
`migration-failure`	The `migrate` command failed
`hook-failure`	A release hook failed
`config-error`	A setting or secret the app needs is wrong or missing
`release-failure`	The release could not be applied
`grant-rejected`	The cluster refused the release, for example because its signature did not verify
`database-unavailable`	The app cannot reach its database
`platform-component`	A Gliss component in the cluster is unhealthy
`app-errors`	The app is running but returning errors
`unknown`	None of the above

A page for each cause will follow as the diagnosis rules ship.