Decreasing the capacity of a replicated file system
Similar to the disk removal in a non-replicated file system, the remove disk option is used to reduce the capacity of a replicated file system.
Before you begin
- To perform this task, you must be a Db2® cluster services administrator.
- Always retrieve the disk name to be removed from the host where the removal command is intended to be run. The reason is because the disk name is unique within the same host only. Do not list the file system to determine the disk name, then run the removal command on a different host.
This task explains how to decrease the capacity of a replicated file system by removing disks from the file system. When the disk is removed from the file system, it is returned to the system and may be reused for other purposes (including being added to another IBM Spectrum Scale or non-IBM Spectrum Scale file system).
About this task
An alert is raised in the following conditions:Disk removal operation can I/O intensive. Therefore, it is recommended to run them at off-peak usage hours.
- If the removal operation leads to unequal size between redundancy group 1 and 2, the alert persists until the size of redundancy groups 1 and 2 becomes equal again.
- If all disks in either redundancy group 1 or 2 are removed, the alert persists until both redundancy groups 1 and 2 exist again and they have the same size.
- If IBM Spectrum Scale detects replication settings of files or directories no longer match with the expected settings after the disk removal operation, a manual replication and potentially rebalance operations are required to correct the settings.
- List the
file system on a host to find the disk name to
be removed from. Using figure 1 as illustration, if M1 is the host
where the disk removal command is to be run, issue the following command
to determine the disk name in the file system
db2cluster -cfs -list -filesystem <fs name>
- On the same host M1, issue the command to remove
disk from the file system. For example, to remove /dev/hdisk6 from
redundancy group 2 of db2logfs file system:
db2cluster -cfs -remove -filesystem db2logfs -disk /dev/hdisk6