Migration of Cisco 9800 WLC 17.3.5a to 17.9.4a

[AnilKumar95946]

We have currently 2 Cisco 9800 on prem CL-Controllers running version 17.3.5. Site A WLC is primary for endpoints in Site A and Site B is secondary WLC for endpoints in Site A and Vice versa. In our current setup we are running on low VM resources (HDD and memory). As per Cisco we can't extend the HDD as we need to rebuild the VMs again. We now built new VMs in 17.9.4a code. We now have migrate all our APs to these new controller. So my question is 

1) What is the best strategy to cutover.? We don't want to change the Controllers IP.

2) How we can minimize the downtime? Site A is acting as secondary for Site B and Vice versa. Should we do the migration on both locations at same day or what should be our approach so that APs do not keep failing over on other WLC. 

3) Please share any document or best practice approach in this regard.

 Thanks

[Leo Laohoo]

Are Site A & Site B a 24x7 site?

[AnilKumar95946]

Yes they are.

[Leo Laohoo]

---

@AnilKumar95946 wrote:

Yes they are.

---

So both sites are 24x7.  Ok, I can work with that. 

1.  Move all APs to WLC A. 
2.  Shut down WLC B. 
3.  Stand up NEW WLC B. 
4.  Move all APs to WLC B. 
5.  Shut down WLC A. 
6.  Stand up NEW WLC A
7.  Move select APs to NEW WLC A.

[Haydn Andrews · Accepted solution]

Given each controller is secondary for the other site:

I would pre-download the 17.9.4a code to existing WLC and pre-download to the APs but not swap/ activate the image

Then build up new VMs

Have the config on them

Shut down Site As WLC

Turn on 17.9.4a WLC for Site A and change IP to be same as old WLC (ensuring it already has the config on it including tags if using static tags)

Fail Site A APs back to it, APs will reboot to change image and experience approx 5-15min outage

Would then fail Site B over and have same outage

Replace Site B WLC Fail back

Now point to note, could split the two but if primary WLC failed all APs would fail over and have to swap image and cause 5-15min outage - but that comes down to the risk appetite 

[AnilKumar95946]

Hi Hayden,

Thanks for your reply. I will test out your suggestion in our lab and will let you know how it goes. 

Much appreciated

[balaji.bandi]

i agree with @Haydn Andrews  advise here. adding, If you are using DNAC - then rolling Upgrade is sensible approach.

yes definably build a new VM (this time plan as per learning lessors of old issue you encountered) 

[marce1000]

 - After upgrade(s) on controller run WirelessAnalyzer  (again) ; use the CLI command show tech wireless ; feed the output into
                                                                                                                                 Wireless Config Analyzer

 M.