Statuses

Instance statuses

The instance's status is its worst component's, in this order: Error, Creating, Updating, Degraded, then Running. An instance is Stopped only when every component is.

Status	Meaning
Creating	A component has not been deployed yet — including a component waiting for its startup wave.
Updating	A component is rolling out a new configuration.
Running	Every component is ready, or stopped while another one runs.
Degraded	A component runs fewer ready pods than it should, and waiting may still fix it.
Error	A component has failed in a way waiting will not fix, or the platform could not apply the instance. It keeps retrying.
Stopped	Every component has 0 replicas.

While the instance is not Running, the Overview shows the startup wave being started as Starting wave — see Startup waves.

Component statuses

The Components tab shows the same statuses per component. A component whose startup wave has not begun yet is Creating, and the tab's description names the wave it waits for. Waiting means the platform has not reported a status for the component yet. A component is in Error when one of these holds, and its row then shows the reason:

Message begins with	Cause
Container is crash-looping	A container keeps exiting; the message has the restart count and the last exit. The row shows it from the first restart, and the component turns Error after the third.
Image pull failed	The image or tag does not exist, or the registry refused the credentials.
The scheduler's reason	No node has room for the pod, or none may run it.
The storage reason	A persistent volume cannot be attached.

Container restarted after OOMKilled is informational: the container ran out of the memory its plan gives it and was restarted, and the status is not changed by it. A bigger compute plan is the remedy — see Scale a component.

Pods

A pod listed on the Components tab is Pending, Running, Succeeded, Failed or Unknown, with Kubernetes' reason when it has one, and a badge with its restart count after its first restart. A component's Ready count counts the pods whose containers are all ready.

Endpoints

Column	Shows
Address	The URL of an ingress endpoint; the address:port of a load balancer endpoint, or Waiting for a load balancer address until the cluster assigns one.
Certificate	Pending certificate until the hostname's certificate is issued, Certificate ready after. A load balancer endpoint has none.

Operations

Creating, updating, starting, stopping and restarting an instance each record an operation on the instance's Operations tab. It succeeds when the instance reaches Running or Stopped and fails when it reaches Error. An instance that reaches none of them within ten minutes fails its operation with Timed out after 600s waiting for '<name>' to reach a stable state (last status: <status>). See Operations.