Unable to login to DXO2 cluster ("Service Unavailable")

Products

• DX Operational Observability

Issue/Introduction

Users are unable to log in or authenticate to the DXO2 cluster. Attempting to access the cluster results in the following JSON error response:

{"error":{"code":0,"message":"Service Unavailable","traceId":"###"}}

Cause

This issue is typically caused by underlying node instability or internal node issues within the on-premises cluster, leading to critical service pods failing or requiring eviction.

Resolution

To resolve this issue, perform the following cluster health checks:

1. Access the Environment: Log in to your DXO2 on-premises environment via your terminal or management console.

2. Check Pod Status: Verify the status of all pods running in the cluster (e.g., using kubectl get pods -A or your equivalent cluster management command).

3. Investigate Node Health: Look for pods that are stuck in a Pending, CrashLoopBackOff, or Evicted state. Note that specific nodes may be experiencing internal issues (such as resource exhaustion or network drops); if so, affected pods may need to be manually evicted or the nodes remediated.

4. Escalate if Necessary: If you verify that all nodes are healthy and all pods are fully up and Running, but the login error persists, please contact Broadcom Support for advanced troubleshooting and assistance.