cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
502
Views
3
Helpful
3
Replies

ISE Upgrade with GUI method

henry.astorga
Level 1
Level 1

Hello ISE Community, I am planning an upgrade of my 3.3 deployment and I was curious if anyone has had success with the GUI upgrade method. In the past I have used the CLI and the backup and restore method and not I am seeing in the 3.5 upgrade guide that the GUI is the recommended method so I am curious to hear of any success (or other) stories and some detail on the time it took to complete. I have a large distributed deployment of 32 total nodes and in the past it has taken an entire weekend starting Friday evening and finishing up Sunday afternoon.

Regards,

Upgrade Fun

3 Replies 3

Arne Bier
VIP
VIP

I have had great success with GUI upgrades since 3.3.  There seems to be a group of people who will never use it because the CLI gives you more control. And that is true - you have 100% control. However, with the GUI it guides you through the process and you can't really mess it up.  I will say however, the one time the GUI caused me some trouble was on a remote PSN that was on a slow WAN.  I was unable to break out of the GUI when I had already started the code distribution. I wanted to push the upgrade bundle to the DISK:/  repo overnight, and then tell the upgrade wizard to fetch the code there.  Now I know to do that in advance and it works great.  The GUI also has the nice feature to specify the upgrade bundle + patch. It applies them both - which is nice.

In my view, if an ISE upgrade fails, it's not necessarily because you used the GUI. I do URT and all the checks beforehand - the rest is up to chance.

Once I got everything set up to use the CLI method and tried it the first time, I can tell you I'll never use the GUI again.  That being said our deployment is rather large, about 30 PSN's across multiple sites.  Each site has 6 PSN's, 2 for TACACS and 4 for RADIUS.  Using the CLI, I can upgrade one node from each site at the same time without disruption (our deployment sits behind load balancers).  This makes it go so much faster.  Using the GUI, I would be here for 6-8 hours babysitting it.  Just my opinion.  

If this posts answers your question or is helpful, please consider rating it and/or marking as answered.

@Christopher Bell you reminded me of another convenient feature of the GUI method - you can also upgrade multiple nodes in parallel. E.g. you can select all you secondaries (e.g. secondary TACACS, RADIUS1/RADIUS2/, pxGrid) and upgrade them all at once. Once those are done and active again, you select all the primary nodes in that cohort.

I don't recall much baby sitting, other than observing the progress bar in the GUI while I got on with other things.