FAQ about disk volumes

This page covers common errors you may encounter when working with disk volumes in Container Service for Kubernetes (ACK), including mount failures, scheduling conflicts, and provisioning errors.

Run kubectl describe pod <pod-name> to retrieve the full error message, then find the matching section below.

Error index

Error message	Situation
Timeout error	Mounting a disk to a node
Zone error	Mounting a disk to an ECS instance
Input/output error	After a system upgrade
`The specified disk is not a portable disk`	Unmounting a disk
`had volume node affinity conflict`	Launching a pod
`can't find disk`	Launching a pod
`disk size is not supported`	Dynamically provisioning a PV

Timeout error when mounting a disk to a node

Cause: Node lacks STS permissions

If the node was manually added to the cluster, it may lack Security Token Service (STS) permissions, which prevents disk mounting.

Solution

Manually assign a Resource Access Management (RAM) role to the instance. For more information, see Bind an instance RAM role.

Zone error when mounting a disk to an ECS instance

Cause: Disk and ECS instance are in different zones

A disk can only be mounted to an Elastic Compute Service (ECS) instance in the same zone within the same region. If the disk and the ECS instance are in different zones, the mount operation fails.

Solution

Move the disk and the ECS instance into the same zone within the same region. To do this, recreate the disk in the zone where the ECS instance is located, or use an ECS instance in the zone where the disk is located.

Input/output error after a system upgrade

Cause: Outdated FlexVolume version

An outdated FlexVolume version can cause input/output errors after a system upgrade.

Solution

Upgrade FlexVolume to V1.9.7-42e8198 or later:

kubectl set image daemonset/flexvolume acs-flexvolume=registry.cn-hangzhou.aliyuncs.com/acs/flexvolume:v1.9.7-42e8198 -n kube-system

Recreate the pod with the error.

For a list of FlexVolume versions and release notes, see FlexVolume.

"The specified disk is not a portable disk" when unmounting a disk

Cause: Disk uses subscription billing

Subscription disks cannot be unmounted. This error occurs if the disk was created with subscription billing, or if the billing method switched to subscription when you upgraded the associated ECS instance.

Solution

Switch the disk's billing method from subscription to pay-as-you-go, then retry the unmount operation.

"had volume node affinity conflict" when launching a pod

Cause: PV and pod nodeaffinity values differ

The pod cannot be scheduled because the nodeaffinity attribute in the persistent volume (PV) configuration differs from the value in the pod configuration. Kubernetes cannot find a node that satisfies both constraints simultaneously.

Solution

Update the nodeaffinity attribute in either the PV or the pod so that both reference the same value.

"can't find disk" when launching a pod

Cause: Invalid or inaccessible disk ID

This error has two common causes:

The DiskID value in the PV configuration is invalid.
The disk belongs to a different account, so the current account does not have permission to access it.

Solution

Check the DiskID value in the PV configuration and correct it. If the disk belongs to another account, use a disk that is owned by the current account.

"disk size is not supported" when dynamically provisioning a PV

Cause: Disk size below minimum

The size specified in the persistent volume claim (PVC) is below the minimum allowed value. Disk size must be at least 20 GiB.

Solution

Update the disk size in the PVC to at least 20 GiB.