Create a Post
Showing results for 
Search instead for 
Did you mean: 

Upgrading to 80.40 and have a dedicated MLM?

Saturday was the great upgrade day to 80.40.

16 hours in, our MDSes, MLM and SmartEvent servers were sporting 80.40.

Things have changed since the early days of upgrades.  When it comes to upgrades, I've always upgraded to the Major and then the Minor on the Primary MDS, then the backup, then the SmartEvent and Log servers in that order.  You do not want to do that with 80.40!

You can no longer upgrade your backup MDS if your primary isn't already on 80.40.  Verifier will prohibit the upgrade on the backup.  Which is super cool.  However, verifier breaks down a little bit on the MLM and SmartEvent servers.  It will not validate version conflicts (major or minor) and will happily allow the upgrade to happen even if there is a mismatch.

ProTip:  Upgrade everything to 80.40 first.  THEN go back around and upgrade everything to the latest Minor. 

Additional Free ProTip:  Make sure you have enough space to take a full image of your boot partition with space left over.  If you don't the installer will fail there as well and revert you back after getting through about half of the import process.

If you don't, your MLM is going to sit there spinning at 58% for about 2 hours (we finally figured this out with TAC's help after the 3rd attempt failed).  Digging into the logs you are not given a clear reason as to why everything is being held up and after 2 hours the system reverts back to the previous version (which it does happily and very well I might add).

Looking at the install guide, this helpful nugget of information isn't there and a relatively new and obscure SK (sk163814) article recently came out about it.

I've made mention of this to some R&D folks and they've promised to make it better.  In the meantime, don't burn your entire day like I did.

TL;DR:  Upgrade everything to 80.40, reboot, make sure everyone is happy then loop through your machines with the latest JHT


0 Kudos
2 Replies

Did try upgrading with a Blink image or just the regular GA image?

0 Kudos

This was just the normal GA image via CPUSE.

Someone had suggested using the Blink image - which I've used on our gateways with good success.  But all of my testing and practice upgrade runs in VM were done with the GA image and that's how my upgrade was planned.

Maybe next time I'll try Blink on the management environment.

0 Kudos