Migrating Data LIFS | NetApp | Page 1

fast kayak Jan 27, 2026, 6:47 PM

#

Question. If you are adding new nodes to a cluster. The old nodes are going away. Let's say a C250 to C80. What really is the recommended methood for either migrating or creating new LIFS? I feel like over time if you created 25 LIFS on the existing cluster, then added the C80 to the existing cluster, why not just migrate the current LIFS? Some think it's better to create new LIFS home on the new node ports, but I feel like to create 25 new LIFS with new IP's is just a waste of time and exhaustion of IPs, if you can simply just migrate the existing. I'm looking for thought, opinions etc. on this. Is the LIF migration process just too flawed to trust with ONTAP?

idle oriole Jan 27, 2026, 7:10 PM

#

I have my view.

If you have done San lifs properly, you have two per node. I would shut down the b/side lifs. Move them to the respective new nodes. Then enable them. When you do any volume moves the clients will always see only one path either on the old nodes or on the new nodes once the move starts. When all volumes are moved, shut down the a/side move them and bring up.
NAS-> if possible relocate volumes on old node 1 to new node 3 and old node two to new node 4. You can then modify NAS lifs whenever, modifying the home-node and if different, the home-port. If auto-revert is enabled, it will automatically re-home the lif. If not you will need to tell ONTAP the send the lif home. Move volumes, deal with the backend indirect io.

When all volumes and lifs are moved, follow the directions to remove the nodes from the cluster

I routinely reuse lifs whenever possible if nodes are being removed. If we’re simply growing the cluster, it’s adding new lifs

You only really need on lif per node per svm and sometimes per protocol.

Meaning: if I’m doing nfs/cifs and I’m not doing jumbo frames, the same lifs may be used. If I’m doing cifs (mtu 1500) and nfs (mtu 9000) id make sure they are on separate ports/tagged vlans. If you happen to have iSCSI or NVMe/TCP or even FCP, those should have two lifs per node, preferably on separate vlans

dreamy sky Jan 27, 2026, 8:41 PM

#

if the old nodes are going away, why would you ever create new LIFs? You will have to reconfigure each and every client. For SAN this is annoying enough but think about CIFS or NFS LIFs, you would need to change that everywhere in AD and/or your Linux hosts

#

I wonder where you got the idea that "migrating LIFs" is a "flawed process"

clear dove Jan 27, 2026, 9:39 PM

#

+1 for moving existing LIFs ✊

Sometimes when adding new nodes I create 1-2 test LIFs on the new nodes to confirm all layer 2 and 3 connections are working fine (especially when much routing is involved and you're connecting the new nodes to other/replacement switches, like a new Cisco ACI or something). Also sometimes customers need some assurance so why not quickly creating some LIFs. But ultimately I always try to reuse the existing LIFs.

idle oriole Jan 27, 2026, 10:37 PM

#

Honestly, If you create the Ports (like tagged vlans) and they magically join the current broadcast-domains, then Layer 2 is good which means Layer 3 (IP) >should< work

#

I always get worried when the L2 doesnt work. Time to go toubleshooting.
I have a customer with not quite ACI, but not straight Nexus (some Nexus dashboard thingy) and they disallow CFM (Connectivity Fault Management) between switches. So when we put one node on one pair of switches and the other node on another set, L2 discovery did not (and still does not) work. They refuse to troubleshoot and do the right thing. Frustrating. ONTAP Layer-2 discovery depends on CFM

clear dove Jan 27, 2026, 10:49 PM

#

ONTAP layer2 port reachability was giving thumbs up.
But the network guy forgot one VLAN on some uplink between old-world and new shiny ACI world.
Fun times 🙃

idle oriole Jan 27, 2026, 10:55 PM

#

And all that VLAN "masking" drives people nuts. ACI Vlan 123 is actually vlan 234 under the sheets...ugh

fast kayak Jan 28, 2026, 4:37 PM

#

dreamy sky I wonder where you got the idea that "migrating LIFs" is a "flawed process"

A 3rd party seems to think we need new LIFS for volume migration rather then move existing.

idle oriole Jan 28, 2026, 4:45 PM

#

Nutso. If you’re migrating NAS to another NetApp cluster: vserver migrate, and it keeps the lifs. If your are upgrading nodes or replacing nodes, then just migrate to new nodes, deal with the possible indirect i/o and be done

clear dove Jan 28, 2026, 4:46 PM

#

Hope your "3rd party" is no NetApp partner...

fast kayak Jan 28, 2026, 4:47 PM

#

That's what I say. Join the new nodes to exsiting cluster. Migrate LIFS and volumes to the new nodes. Decommission the old nodes. Simple

#Migrating Data LIFS