As others suggested:
Raise a ticket with NetApp, if this is affecting user experience.
As you said, the SnapMirror to other ONTAP had no issues, It could be just a case where 'destination' volume was still mapping the logical blocks to physical blocks as part of deswizzle process. It depends on the load on the Node (Aggregate) where the dest_volume was hosted.
You might have just ran an update, broken mirror and mounted R/W.
Steps to isolate this:
::> set d
::*> system node run -node <node_name> -command "wafl scan status"
node_name = where the aggregate is hosting the destination volume (NFS).
Output = If there is none for that volume, then :
1) Unmount filer vol from redhat
2) Re-mount it on redhat
3) Try acccessing again.
If it still behaves that way , use 7mtt.
Give it a try.