Soon I've got to add more RAM to our 3 ESXI hosts, however I've never had to shut down these at once before, does anyone have a good source for an order of operation to performing such maintenance.
Everything depends on the following:
- are you hosts managed by vcenter?
- are they in DRS cluster? if yes what is an automation level?
- are VMs on shared datastore or not?
Generally speaking if you have fully automated DRS cluster and all VMs are located on shared datastore you can just enter one host into maintenance mode > shutdown it > add RAM > power on it > exit host from maintenance mode.
Than repeat the same actions with other hosts.
You have to do this one by one ESXi host .
You have to move the VMs from one host to other two nodes distributed . ( In case you have don't have resources - RAM/CPU on ESXI you have to shutdown less priority vms and move )
After moving all VMS to other node , put this host to maintenance mode .
Now you can shutdown the host and upgrade RAM
Not in a DRS cluster, VMs are split through multiple datastores through 2 SANS (VSANS)
Hosts are managed by Vcenter, however Vcenter also exists on one of these hosts.
(There is not enough resource to move all VMs from on over to another.)
If you have vSAN you have to enter maintenance mode host before shutdown.
When you enter host into maintenance mode you have option to evacuate data from host to keep protection level of VMs on VSAN cluster or to leave data on host and reduce protection.
If don't have enough resources on other hosts in cluster to move VM then you have to power them off manually.
vcenter VM you can move with vmotion, this will allow you to have single management console of all hosts.
If there is no enough resource may be you have configured HD admission control - you can disable temporary .
Other as i said earlier you have to shutdown less priory vms and proceed upgrade .