Issue is that starting an application (for testing) in many batch sessions will end up in ERROR that is typical when access to its temp space is restricted and the application could not start

Example:
start session 1 via command line a batch session > starts correctly
end session
start session 2 via command line a batch session > end in ERROR typically reported by application when it could not access its temp space (detail below)

Behavior: Sometimes in round robin manner 1 yes 1 no, sometimes 8 session start without error and 9th errors.

The application choses one of the 3 temp spaces (listed below) in a random order (application internal configuration which is working).

Two nodes each have three 250GB single disks.

1. Fdisk /dev/sdb , sdc, and sdd with a single partition
2. Created a volume group on each partition, WORK1, WORK2, WORK3
3. Created a logical volume, WORK1, WORK2, AND WORK3 and at the same time formatted with XFS using LVM gui tool

All the disks are single disk, no striping from logical volume level.

Enterprise storage and no barrier used.

Any insight is deeply appreciated.