Skip to main content

Troubleshooting

Collect diagnostics

kubectl get all --namespace <namespace>
kubectl get pvc --namespace <namespace>
kubectl get events --namespace <namespace> --sort-by=.lastTimestamp
kubectl logs <pod-name> --namespace <namespace> --all-containers=true

For operators:

kubectl get crd | grep -E 'anzo|anzograph'
kubectl get anzo --namespace <namespace>
kubectl get anzograph --namespace <namespace>
kubectl describe anzo <name> --namespace <namespace>
kubectl describe anzograph <name> --namespace <namespace>

For Helm:

helm status <release-name> --namespace <namespace>
helm get values <release-name> --namespace <namespace>

Common issues

SymptomLikely causeCheck
ImagePullBackOffWrong image tag, registry access, or pull secret.Pod events and image values.
Pod stuck PendingInsufficient resources or unresolved PVC.Node capacity and PVC events.
CrashLoopBackOffMissing secret, invalid license, invalid config, or resource limits.Container logs and events.
Custom resource does not reconcileOperator not ready, CRD mismatch, RBAC issue, or wrong namespace watch.Operator logs and CR description.
Service unreachableIngress, TLS, load balancer, source range, or network policy issue.Services, ingress, events, and endpoint health.
Graph Lakehouse data missing after upgradePVCs not retained or database reinitialized.Upgrade procedure, PVC history, and backups.

Redact secrets, tokens, passwords, license strings, and private keys before sharing diagnostics.