I recently went through migrating a production DNAC cluster running on Gen1 appliances (UCS C220 M4s) running DNAC 2.1.2.5 to Gen 2 servers (UCS C220 M5s) and want to share the process and a few tips. Here is an overview of the steps taken in order to ensure the migration was successful:
-Backup config from old cluster which will get used for restore later
-Download production version ISO & create bootable stick
-Reimage all 3 new servers to the same running version
-Configure CIMC on all 3 new nodes
-*Upgrade CIMC firmware if necessary
-Begin configuring G2 Server 1 by starting a new cluster
NOTE: You cannot change the cluster link, cluster link IP, or any VIP on any configured/used interfaces once done using the installer so make sure you have a plan for this (if you goof here and will want to change the cluster link you will need to reimage).
![dnac_maglev_cluster_ip_error.PNG dnac_maglev_cluster_ip_error.PNG](https://community.cisco.com/t5/image/serverpage/image-id/106438i4EF660165B20295D/image-size/medium?v=v2&px=400)
The other interfaces & IPs can be changed later to mirror production IPs. Lastly, you cannot use the same subnet on multiple interfaces.
-Continue building out new cluster by configuring/joining the other 2 to G2 Server 1
-Once the other 2 nodes are joined login to the web UI and enable HA in system360 (per TAC this needs to be done prior to performing restore)
-Download and install apps to ensure the versioning matches production servers (per TAC they need to be spot on for restore operation) ($ maglev package status)
-Once the G2 servers are joined in a cluster with HA enabled & all apps match the G1 server apps, attempt restore from backup.
NOTE: I actually used winscp to move the backup tar file onto one of the nodes in the new cluster, pointed the backup config directory there, and performed the restore without issue. Be aware there is a known bug (which we did not hit) for 2.1.2.5 restore operations see: Bug Search (cisco.com)
-After successful restore, do a few spot checks (make sure inventory is good, etc.)
Cutover/Migration steps:
-Shutdown all servers in old cluster ($shutdown -h now)
-Reconfigure new cluster enterprise/mgmt interfaces to mirror production IPs using maglev wizard ($sudo maglev-config update)
-Once that is done move to inventory, let devices sync up and get happy
-At this point the cutover is essentially done. Test provisioning devices, pxgrid connection to ise, & generate new TLS cert if necessary.
Few extra things to be aware of: After install on first UI login you are prompted to change the admin password again. -If you need to change it back to what it originally was (what you first wanted the pw to be) use: $ sudo magctl user password update admin -p <pass> TNT0:
![dnac_2125_admin_change.PNG dnac_2125_admin_change.PNG](https://community.cisco.com/t5/image/serverpage/image-id/106435i9A5B5D1968785665/image-size/medium?v=v2&px=400)
-If you do not want to use the KVM to use the wizard there is a new web installer option which is really cool and aides in ensuring you are configuring the right interfaces:
![dnac_ui_webinstall_LI.jpg dnac_ui_webinstall_LI.jpg](https://community.cisco.com/t5/image/serverpage/image-id/106436iCB0C5EBBC33749CF/image-size/medium?v=v2&px=400)
I would definitely recommend engaging TAC as well so that you have their support in case encountering any bugs. Good luck & HTH!