Container Service for Kubernetes:Manage the csi-plugin and csi-provisioner components

The csi-plugin container reports:

exec /usr/bin/plugin.csi.alibabacloud.com: exec format error

Cause: csi-plugin supports both amd64 and arm64 architectures. This error occurs when an image pull is incomplete -- the image metadata exists, but the binary is invalid. A forced node shutdown during the image pull is a common trigger. Check the ActionTrail logs for the ECS instance to confirm whether a shutdown command was executed.

Solution:

• Option A: Scale out the cluster by adding a new node, then drain the current node.

• Option B: If adding a new node is not possible:

  1. Drain all applications from the current node, then remove the node from the cluster.

  2. Log on to the node and delete all containers, if any.

  3. Delete all files in the /var/lib/containerd directory.

  4. Add the node back to the cluster.

The sidecar container in the csi-provisioner pod caches information about pods, persistent volumes (PVs), and persistent volume claims (PVCs). As the cluster grows, this cache can cause OOM errors.

• Managed version: Submit a ticket for assistance.

• Self-managed version: Adjust the memory limit based on cluster size:

  1. Log on to the ACK console. In the left navigation pane, click Clusters.

  2. On the Clusters page, find the target cluster and click its name. In the left navigation pane, click Add-ons.

  3. On the Add-ons page, find csi-provisioner, click the icon, and select View YAML.

  4. Edit the YAML file to increase the memory limit.

Pod monitoring shows unusually high network traffic for csi-plugin.

Cause: csi-plugin mounts NAS volumes on nodes. When a pod uses a NAS volume, the NAS request traffic from that pod passes through the csi-plugin namespace. Cluster monitoring records this traffic against the csi-plugin pod.

No action required. The traffic is only recorded by monitoring. It is not duplicated and does not consume extra bandwidth.

Running kubectl logs csi-provisioner-xxxx -n kube-system shows:

failed to renew lease xxx timed out waiting for the condition

Cause: csi-provisioner is a high availability (HA) component that runs multiple replicas. The pods use the Kubernetes Lease object for leader election. The pod that acquires the lease becomes the leader and serves the cluster. This error means a pod failed to reach the API server.

Solution: Check whether the cluster network and the API server are functioning normally. If the issue persists, submit a ticket for assistance.