Troubleshoot Kubernetes

Use this section to resolve common issues with E2E Kubernetes clusters - both portal/cluster-level problems and in-cluster workload problems.

---

Cluster and Portal Issues

Symptom	Likely cause and fix
Cluster creation fails or hangs	A security group missing TCP 6443 / UDP 8472, insufficient credits, or a temporarily unavailable plan. Check the attached security group and your credit balance, then retry.
kubectl cannot connect	The attached security group must allow TCP port 6443. Confirm the kubeconfig is current - see Connect to a Cluster.
Kubernetes option not offered in a region	Kubernetes is not available in the Mumbai location. Use a supported region.
Master plan upgrade is disabled or blocked	The cluster must be Running, not Creating/Terminating, and you must have enough credits. Only higher-tier plans in the same series are offered. See Actions.
Cannot delete a worker node	You cannot go below the pool minimum (2 for compute, 1 for GPU). Delete the whole pool instead. See Node Pools.
SSL certificate about to expire	Renew from the Cluster Details tab, or let it auto-renew.
A tab shows "please wait"	The cluster is Creating or Upgrading Master. Most tabs become usable once it is Running.

---

Workload Issues

In-cluster workload problems and their fixes:

Issue	Guide
Pods crash and restart repeatedly	CrashLoopBackOff
Image cannot be pulled	ImagePullBackOff
A node shows NotReady	Node Not Ready
Pods killed for memory	OOMKilled
Pods stay Pending / cannot schedule	Pod Unschedulable
PVC stuck in Pending	PVC Pending
Istio resources not creating	Istio Not Creating

---

Related Resources

Resource	Use it for
Manage Kubernetes	The full management surface.
Connect to a Cluster	kubeconfig and access.
Kubernetes Guides	In-cluster how-to guides.