GPFS upgrade hangs when using "Check for updates" option on vsys instance page

Answer

Solution:

Customer had to set automount to true on both server and client nodes (each VM needed to set automount to true):

1. You can check whether file systems are set to automount by using the GPFSManager's "get cluster status" operation

If any file systems are not set to automount, you can change that with the following command:

su - gpfsprod -c 'sudo /usr/lpp/mmfs/bin/mmremotefs update fileSystemName -A yes'

2. After this property was set, the instances could be upgraded smoothly. Note: As the number of VMs in the instance increases, so does the time it takes for this process to complete. There are several pieces of software being upgraded, not just GPFS, so plan for a longer maintenance window for larger instances.

Notes:

Here are some notes from the GPFS pattern team regarding this issue:

The file systems created on GPFS server after deployment of client instance need to be defined as remote GPFS file system with enabled auto mount option. They also need to be manually mounted on client instance. We need to set auto mount option explicitly true using the following 'mmremotefs' command

Command to define a new remote GPFS file system with enabled automount option:

su - gpfsprod -c 'sudo /usr/lpp/mmfs/bin/mmremotefs add <yourfileSystemName> -f <yourfileSystemName> -C <yourclustername> -A yes -T /gpfs/yourfileSystemName'

Here, <yourfileSystemName> refers to name of file system on remote cluster, <yourclustername> refers to remote cluster name, and -A yes defines automount option and is mandatory here.

Command to mount the specified file system manually:

su - gpfsprod -c 'sudo /usr/lpp/mmfs/bin/mmmount <yourfileSystemName> -a'