After you deploy a container group on SaladCloud , some nodes may experience errors. To detect and diagnose these errors, SaladCloud provides an API endpoint here to access high-level details of recent errors that occurred on a workload. In addition, you may view and download a log of these errors from the portal by clicking the ‘Workload Errors’ tab on the container group details page. Note that this feature, whether accessed from the portal or the API, only returns up to the 50 most recent errors for a workload which occurred in the last hour.

Fields

detail - An error message and the exit code from the container. Possible error details are detailed below.

failedAt - The UTC timestamp when the error occurred.

machine_id - The unique ID of the node on which the error occurred.

allocatedAt - The UTC timestamp when your container was allocated to the node.

startedAt - The UTC timestamp when your container was started on the node. If this field is null, the container failed before reaching a started state. You can learn more about different container states here.

Error Details

These are the possible details which can be returned, what they mean, and what you can do to fix them.

DetailCauseHow to fix
RunFailureThe container exited after it started.This indicates a likely issue with the container itself. If you have configured an external logging service, investigate the logs returned from the container to diagnose the issue.
StartFailureThe container failed to start.This indicates either an issue while starting the container itself or with the node. If a command was provided, check it can be executed successfully in the container. If start failures are occurring regularly, contact SaladCloud support.
StartupProbeFailureThe configured startup probe failed.Check startup probe parameters. Ensure that you are providing enough time for the container to start before failing the probe.
LivenessProbeFailureThe configured liveness probe failed.Check liveness probe parameters. Ensure that you are providing enough time for the probed application to reach the desired state before failing the probe.