We had a working setup running Oracle DNFS on a 7-mode FC MetroCluster. After transistion to C-Dot with ONTAP 9.5 DNFS produces strange behavior, slowness and Oracle errors. We don't find any reason. At the end we disabled DNFS and switched to NFS. We search for anyone who has a working C-Dot setup with DNFS to compare configurations (perhaps we missed something).
Been a while but if I recall you want to have at least one network interface (LIF) per controller and the oracle dnfs config file should contain all those IP addresses.
also make sure
your export policy rules are correct
disable udp on the nfs vserver
enable 64-bit file handles
make sure flow control is off at the netapp ports and the switch and it would not hurt to employ on the host also
use jumbo frames if possible and check they are properly enabled end to end! People miss this so check!
here is the odd part
of you are using an ifgrp and Oracle is the only thing accessing you may want to break into access ports and define one LIF per interface instead and update Oracle. It will use all paths much more efficiently!
I don't know much about Oracle but hopefully that points you in the right direction. Support doesn't have resources to troubleshoot Oracle as we only handle the NetApp side of things. I'd check the TRs, maybe the account team, but Oracle proper can really help the best.