Troubleshooting database cluster issues
If your database cluster isn’t behaving as expected, start with these checks:
-
Check cluster status.
kubectl describe databasecluster <cluster-name> -n <your namespace> -
Check the operator logs.
kubectl logs -n <operator-namespace> deployment/everest-operator-controller-manager
The table below summarizes frequent issues, their possible causes, and how to resolve them.
| Issue | Possible Cause | How to Check / Fix | Resolution |
|---|---|---|---|
| Cluster stuck in creating or initializing | Insufficient CPU, memory, or nodes available | - Inspect pod events:kubectl describe pod <pod-name> -n <namespace>kubectl describe pod -l app.kubernetes.io/instance=<db-cluster-name> -n <namespace>- Check node resource allocation: `kubectl describe nodes | grep -A5 "Allocated resources"<br /><br />- Monitor live usage:<br />kubectl top nodes<br />kubectl top pods -A` |
| Cluster in error state | Misconfigured storage class or insufficient resources | - Check available storage classes:kubectl get storageclass- Inspect pod events: kubectl describe pod <pod-name> -n <namespace>kubectl describe pod -l app.kubernetes.io/instance=<db-cluster-name> -n <namespace>- Check node resource allocation: kubectl describe nodes | Update the storage class in the cluster spec or request more resources. |
| Backup or restore fails | Invalid backupStorageName or missing/incorrect S3 credentials | - Inspect backup:kubectl describe dbbackup <backup-name> -n <namespace>- Verify secrets: kubectl get secrets -n <namespace> | Fix the BackupStorage configuration and update S3 credentials secret. |
| Monitoring not working | Missing or invalid PMM credentials | - Check monitoring config:kubectl get monitoringconfig -n <namespace>kubectl describe monitoringconfig <name> -n <namespace> | Recreate the PMM secret with valid base64-encoded credentials. |
| Proxy not starting | Wrong proxy type or insufficient resources | - Check proxy pod status:kubectl get pods -n <namespace>kubectl describe pod <proxy-pod> -n <namespace> | Correct the proxy type in spec and/or increase CPU/memory requests. |