Managed Service for Kubernetes node group hosts stay in the RECONCILING status for a while

Issue descriptionIssue description

Some (or all) VMs of your Managed Service for Kubernetes node group remain in the RECONCILING status for more than 20 minutes.

Diagnostics and issue reproductionDiagnostics and issue reproduction

• Check your quota consumption for Managed Service for Kubernetes. If the consumption level is approaching the current quotas, request a quota increase. For this, you must have the quota-manager.requestOperator role or higher, such as editor or admin.
• If the quotas are not exceeded, check the state of VMs belonging to the node group through this command:

yc managed-kubernetes node-group list-nodes <node_group_id> --format yaml

If the command output includes the Kubelet stopped posting node status message, it means the VM belonging to the node group experienced a failure in the OS or application services that ensure operability of this host within the node group.

SolutionSolution

Follow these steps to fix the issue:

1. Connect to the VM in question over SSH. You can learn how to do it in this Managed Service for Kubernetes guide.
2. Establish an interactive session with the problematic VM and run the following commands:

   sudo service containerd restart
   sudo service kubelet restart
   

service containerd status
service kubelet status

Alternatively, you can restart your VM from the management console or YC CLI.