Shutting down an IBM Storage Scale cluster

You might need to shut down an IBM Storage Scale cluster in an emergency.

Stop the protocol services on all protocol nodes in the cluster by using the mmces service stop command.
For example:
```
mmces service stop nfs -a
mmces service stop smb -a
mmces service stop s3 -a
```
Unmount all file systems, except the CES shared root file system, on all nodes in the cluster by using the mmumount command.
Stop GPFS daemons on all protocol nodes in the cluster by using the mmshutdown -N cesNodes command.
Unmount all file systems on all nodes in the cluster by using the mmumount all -a command.
Stop GPFS daemons on all nodes in the cluster by using the mmshutdown -a command.

After performing these steps, depending on your operating system, shut down your servers.

Before shutting down and powering up your servers, consider the following points:

You must shut down NSD servers before the storage subsystem. While powering up, the storage subsystem must be online before NSD servers are up so that LUNs are visible to them.
In a power-on scenario, verify that all network and storage subsystems are fully operational before bringing up any IBM Storage Scale nodes.
On the Power® platform, you must shut down operating systems for LPARs first and then power off servers by using Hardware Management Console (HMC). HMC must be the last to be shut down and the first to be powered up.
It is preferable to shut down your Ethernet and InfiniBand switches using the management console instead of powering them off. In any case, network infrastructure such as switches or extenders must be powered off last.
After starting up again, verify that functions, such as AFM and policies are operational. You might need to manually restart some functions.
There are a number other GPFS functions that could be interrupted by a shutdown. Ensure that you understand what else might need to be verified depending on your environment.