Troubleshooting

An error from rft, and each finding in a support bundle's analysis, starts with an id from this page. Find the id below for what it means and how to fix it.

To send RigFabTech the cluster's state, collect a support bundle.

RFT-E-ROLLOUT

A workload of the base chart or a product has fewer ready pods than it should.

• Run kubectl describe and kubectl logs on the pod the support bundle names.
• For a product, kubectl get platformapps --all-namespaces shows what the controller waits for.

RFT-E-NODE-NOT-READY

A node's kubelet stopped reporting, and Kubernetes starts nothing new on that node. Bring the node back, or drain and replace it.

RFT-E-DISK

A node reports disk pressure, and Kubernetes evicts pods from it. Free space on the node or grow its disk.

RFT-E-IMAGE

A pod can't pull its image.

1. Write the pull Secret again with a current pull token:

   rft login --username <SITE_ID> --password-stdin --pull-secret rft-system/rigfab-pull < pull-token.txt

2. Run rft check. It tests that this machine reaches the registry.

3. Check that the nodes reach the registry too.

RFT-E-CAPACITY

Kubernetes can't place a pod because the nodes lack CPU or memory. Add capacity, or uninstall a product.

RFT-E-LICENCE-CHECK-IN

The licence agent hasn't checked in with the licence service for 2 days. A missed check-in never turns a product read-only. After 7 days without one, admins see a banner in every product.

1. Check that the cluster reaches the licence service on HTTPS.

2. Read the agent's log:

   kubectl --namespace rft-system logs deployment/rft-base