Replacing node disks

When a disk fails, data replication begins immediately and automatically, and finishes when the disk becomes healthy again. This might take some time depending on the cluster configuration. In case of a disk failure, the cluster administrator should determine whether this failure is due to a temporary or permanent condition:

  • If the condition is temporary (for example, a physical disconnection), you can try keeping a failed disk in the cluster and avoid replacing it. After fixing the issue, you can mark the disk as healthy. If successful, the cluster can be considered to be in a normal operating state.
  • If the condition is permanent, it is recommended to release and replace the disk as soon as possible. A failed disk can potentially lead to several issues, especially if the cluster does not have enough available resources, such as degraded performance, lack of space, inability to restore data redundancy, and other.

To replace a disk with a new disk, you need to release the old one from the storage cluster. If the new disk contains any data, the disk will not be considered suitable for use in the storage cluster.

After the replacement, you will need to assign the role of the released disk to the new one:

  • The Storage role will be assigned automatically if you enabled automatic configuration of new disks before a disk failure.
  • All other roles need to be assigned by hand.