linux OS mounting using path name instead of powerpath pseudo name

We had an issue in the past few months with a new server running rhel 7.9 and powerpath 7.2 using FC attached storage on a Unity 500. the application would stop working and complain of database corruption. When i looked at the file systems, instead of being mounted on the powerpath pseudo name like /dev/emcpowera, they were mounted on one of the path names like /dev/sdb

After running fsck and rebooting, this would fix the problem and the mounts were using the pseudo name again.

We now think this occurs after adding a new lun and running mount -a. Here is the full email from my linux admin. Are we doing anything wrong? I was thinking we refernce the powerpath name directly in ftsab instead of using uuid.

It looks like the process of creating and mounting a new file system triggers the issue.

 

Everything looked good after Al added a new LUN.

 

Then I ran the following commands to create a new file system:

 

mkdir /tsminst1/TSMcont-encrypted08   ---- to create a directory as a new mount  point

mkfs -t xfs -K /dev/emcpowert    --- to create a new file system and format it as XFS

xfs_admin -L TSMcont-enc8 /dev/emcpowert   ---- to assign a label (optional)

blkid -o list|grep emcpower   --- to see what is UUID for a new file system

vi /etc/fstab   --- to add new file system (with UUID) for automatic mount after reboot

 

 

mount -a   -  This reads all file systems from /etc/fstab file and tries to mount it.

 

Usually it works and does not disrupt already mounted file systems, but this time it has caused two file systems to get re-mounted on one path instead of emcpower device.

 

So next time I will have to mount the new file system individually and avoid mount -a command.

 

May be we can do some capacity planning and create all file systems for the whole year to

 

Let’s stop TSM and reboot again to fix the issue before we will start seeing corruptions.