Last Updated: Jan 15, 2026
Introduction
This article addresses the "No agent available" error encountered when attempting to fetch pod logs or perform kubectl exec commands within a Crusoe Managed Kubernetes (CMK) cluster.
When a cluster is unable to communicate with the control plane, users may experience the following symptoms:
kubectl logs <pod_name>fails with the error:Error from server: Get "https://<IP>:10250/...": No agent available.kubectl execcommands fail to connect to pods.Admission webhooks (e.g., Kyverno) may fail with similar connection errors.
This issue typically occurs when the Konnectivity agent - the component responsible for proxying traffic from CMK control plane to the worker nodes - cannot establish a connection back to the cluster endpoint.
Prerequisites
- An up and running CMK cluster.
Step-by-Step Instructions
Verify firewall rules
Make sure that you have this rule, which is created by CMK when you create your cluster. Refer to this documentation for additional details.
crusoe networking vpc-firewall-rules list |grep konnectivity name id direction protocols sources source ports destinations destination ports ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ cmk-cp-konnectivity-cp-25d0b486 2b191f10-a669-4b03-bfe5-fd2e95de7ee1 ingress tcp ,, 8132,8133,8134
Also, make sure that all outbound traffic can flow from the VPC subnet where you are running your nodepool VMs back to the control plane.
It is a common practise to enable all outgoing traffic to 0.0.0.0/0
crusoe networking vpc-firewall-rules list name id direction protocols sources source ports destinations destination ports ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ default-allow-tcp-udp-egress e2deef61-2d1a-45ca-ae20-6c82fae1c685 egress tcp,udp 172.27.0.0/16 1-65535 0.0.0.0/0 1-65535
Verify Konnectivity Agent Status
Check if the Konnectivity agent pods are running in the kube-system namespace.
kubectl get pods -n kube-system -l k8s-app=konnectivity-agentIf the pods are running but the error persists, proceed to check their logs for connection failures.
Inspect Konnectivity Agent Logs
Examine the logs of a Konnectivity agent pod to understand what might be happening when communicating with the control plane, and why the error No agent available is thrown.
kubectl logs -n kube-system <konnectivity-agent-pod-name>The command might not work as this is the issue we are trying to resolve. You can extract the pod logs from the worker node under /var/log/pods/kube-system_konnectivity-agent-<>.
Look for any errors in the logs of the konnectivity agent.
Example error you might see if the configuration of the konnectivity agent is not correct:
rpc error: code = Unavailable desc = connection error: desc = \"transport: Error while dialing: dial tcp: lookup <cluster-address>-internal.us-east1-a.cmk.crusoecloudcompute.com on 10.233.***.***:53: no such host\"
Verify the Konnectivity Agent settings
In konnectivy-agent configuration make sure you have the following:
kubectl describe deployment konnectivity-agent -n kube-system
-
-
Command:
/proxy-agent
Args:
--logtostderr=true
--ca-cert=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt
--proxy-server-host= <Cluster DNS name>
--proxy-server-port=8132
--admin-server-port=8133
--health-server-port=8134
--service-account-token-path=/var/run/secrets/tokens/konnectivity-agent-token
-
-
-
Note: Replace <Cluster DNS name> with the DNS name of your CMK cluster. You can obtain the DNS name using this command:
crusoe kubernetes clusters list --json | jq -r '.[].dns_name'
If the above doesn't resolve the issue, please contact Crusoe Support.
Additional Resources