1. Yes, I do this: I suspend and resume jobs, it works.
But there is one serious problem: when this method is used for backups of VMs, the backup of a specific VM
is not suspended, but is restarted, therefore this method is not so acceptible for jobs that back up of VMs with very large vdisks.
2. Let me tell you what kind of problem I have in mind when I ask about this possibility of forced
switching of mount paths for the active job.
I have 2 MAs with a shared disk library that has 8 mount paths. Half of them are the preferred mount
paths for MA #1, the other half are preferred for MA #2. I made a “classic” GridStor configuration with
two storage policies , as in example No. 1 here:
http://documentation.commvault.com/commvault/v11_sp20/article?p=10870.htm. My policies are both under the same global ddup policy.
In this way, jobs are distributed between MA #1 and MA #2 and places the copy streams on the
appropriate mount paths, which are preferred for MA, which is used in the default data path of the
storage policy. But sometimes the distribution of streams between mount paths is not optimal.
For example, a virtual machine has 2 v-disks, the job creates 2 streams and for some reason places
them together on one mount path, and does not distribute them along 2 mount paths, while other
mount paths (which are the preferred for this MA ) are free. The "Spill and fill mount paths" mode
is configured for this library, of course.