How can I debug ReadOnlyLocalSSDDetected?

Google Cloud Serverless Applications
1

We have a GKE autopilot cluster, which seems to behave… Strange. I’m not sure why.

Our pods get evicted quite often. For example, we have a redis-master node running. From the pod logs, I can only see Received SIGTERM scheduling shutdown... If I look at further down the logs: core/v1/namespaces/redis/pods/redis-master-0/eviction. For some reason the pod was evicted.

If I then look at the node logs, I can see this: ReadOnlyLocalSSDDetected a few seconds before. After that all pods on the node are evicted.

I’m not sure, but I think this is the reason. Any tips how to go forward from this?

Running version: 1.32.6-gke.1060000

2

Hi @Gustav_Elmgren,

Looking into Internal documentation, It seems that this is a known issue and needs further assistance. Please contact Google Cloud Support for further investigation.

3

Hi @francislouie,

Thank you for the response. Do we need to pay to further investigate this issue? We only have basic support, and according to the link you sent, we need at least standard support to contact Google Cloud Support?