Staged Upgrades with Stacks Rebooting At the Same Time

[Slobs2]

As the title would suggest, I'm looking for everyone's experience and info on the behavior of the staged upgrade tool with stacks. I was told a few times by support that the best practice for using the staged upgrade tool is to add all members of the stack to be upgraded to the same window of time. I was also told that after they download the update, there is a "back-off" timer that should prevent them from all rebooting at the same time. I am not finding this to be the case after two stacks I have updated, and in both cases all member rebooted at the same time. I was also told to not try to stagger the updates of switch stack members as this might cause issues with the stack. What is everyone's experience with this scenario? I am updating from 10.26 to 10.45. Any insight is much appreciated!

[PhilipDAth · Accepted solution]

If you select any switch in a stack to upgrade then all members will upgrade, and all members will reboot at the same time.

It is not possible to upgrade only one member of a stack only and it is not possible to have them reboot one at a time.

This article talks about this process.

https://documentation.meraki.com/Architectures_and_Best_Practices/Cisco_Meraki_Best_Practice_Design/...

[Slobs2]

Thanks @PhilipDAth for your response. I did see that documentation as well, I guess where it says "rebooting close to the same time" was just ambiguous to me. Thanks! 

[PhilipDAth]

Let's phrase it this say, "rebooting close to the same time" means you will experience a 100% outage of the stacking.

[Slobs2]

@PhilipDAth Haha yes that would be much more accurate!

[Slobs2]

@PhilipDAth One last question. If I complete all the edge switch stack upgrades using the staged upgrade tool then only have my two MS425 aggregation switches left to update (which is where I really would like redundancy to occur), does using the regular firmware upgrade page under Org->Monitor behave any differently?

[PhilipDAth]

Well, probably not.  Take a read of this:

https://documentation.meraki.com/Architectures_and_Best_Practices/Cisco_Meraki_Best_Practice_Design/...

And not this bit in particular:

"The switch starts a countdown of 20 minutes to allow any other switches downstream to finish their download"

So when you trigger an upgrade of all switches in a network at the same time, no switch will reboot until all of the switches have complete downloading their new firmware - or 20 minutes have passed (which usually means an upgrade failed).