The safest approach is to just follow the procedure outlined in your upgrade advisor report:
(a) Determine the destination path for each SnapMirror relationship using the command: snapmirror show (b) For each destination volume, enter the following command to allow existing SnapMirror transfers to finish. snapmirror quiesce -destination-path destination If there are no active transfers for the SnapMirror relationship, this command sets its status to Quiesced . If the relationship has active transfers, the status is set to Quiescing until thetransfer is completed, and then the status becomes Quiesced (c) Verify that all SnapMirror relationships are quiesced: snapmirror show -status !Quiesced This command displays any SnapMirror relationships that are not quiesced. (d) If any SnapMirror relationships are currently transferring, do one of the following options: (i) Wait for the transfers to complete before performing the Data ONTAP upgrade. (ii) Stop the transfers by entering the following command: snapmirror abort -destination-path destination -h
If you have dedupes running, I'd issue a:
volume efficiency show -vserver * -op-status Active
And then just issue a stop command on any active processes. Deduplication is opportunistic and will chew through CPU and such which isn't necessarily a good thing during an NDU.
Hope that helps,
(BTW - the previous comment about going to 9.1P2 is spot on - lots of good stuff in 9+)