Core/Server Switch Upgrade - best practice

[This2shallpass]

Hi, 

I will be updating the firmware on our Meraki switches soon. This will require rebooting of switches. Should I shut down all VMs as best practice or it is not required? Will the VMs running during the upgrade procedure freeze & pickup automatically from where they left off once online or they can go corrupt.

Thanks

[Brash · Accepted solution]

This question is ultimately dependant on your network design and risk tolerance.

A few points of note.

If your hosts have uplinks to to 2 separate switches, they will reboot alternately and therefore the hosts and VM's should see minimal impact.

This assumes your hypervisor pathing is set up correctly.

If you have the VM's storage traffic (iSCSI, NFS etc) going over the network. Most hypervisors can handle a small delay in connectivity to the VM storage, but won't handle a larger outage time. It's always a more enjoyable experience to shutdown critical VM's prior and bring them up again later than to have to recover corrupt VM's.

[alemabrahao]

As for vMware, I don't think it's a problem to leave the server on, however, I've seen cases of applications that didn't behave well during a switch reboot in general (regardless of whether it's Meraki, Catalyst, HP, etc).

[This2shallpass]

Thank you @Brash @alemabrahao for your viewpoint

Shutting down the VMs is the best option.