2.3. Releasing Nodes from Cluster¶
To release a node means to remove it from the cluster (e.g., for maintenance). As the node may be running services needed by the cluster, do the following prior to releasing it to avoid cluster degradation:
If the node runs one of the five required metadata services, add a metadata role to another node. You need to make sure that the cluster has at least five metadata services running at any time.
If the node has any access points, make sure that the same access points are configured on other nodes in the cluster as well.
If the node has iSCSI targets, move them to a different node.
If the node has an S3 gateway, reconfigure DNS for the S3 access point to remove the node from DNS records. Next, release the node from S3 in the corresponded sections of the SERVICES screen.
Make sure the cluster has enough storage space to accommodate the data from the released node.
Once you initiate the release, the cluster will start replicating data chunks that were stored on the released node and distributing them among other storage nodes in the cluster. Depending on the amount of data to replicate, the process may take as much as several hours.
If necessary, you can also release a node forcibly, that is, without replication.
Warning
Releasing nodes forcibly may result in data loss.
To release a node from a cluster, do the following:
On the NODES screen, click the node to release.
On the node overview screen, click Release.
If necessary, in the Release node window, check force to release the node forcibly (highly not recommended).
Click Yes. The released node will return to the UNASSIGNED list on the NODES screen.