Upgrade Host Clusters

[rajeevsrikant]

Regarding Upgrade Host Clusters I have the below query.

Cluster A - 5 hosts - DRS Enabled

In each of the 5 hosts I have one VM which is the storage VM.

This VM can not be vMotion to the other hosts.

The option I have is i need to shutdown this VM , since this VM can not be vMotion & because of this Host upgrade will not start.

I can shutdown the VM, but how will i know which host is selected 1st for the upgrade & what will be the sequence.

So what is the best option for this.

My understanding is when I start the Click Upgrade Available option for this cluster, randomly 1 host will be selected & DRS will try to move all the VMs to the other hosts.

Only this VM can not be vMotion & it will get stuck. So that time I can login to that particular VM & do graceful shutdown.

Once upgrade for this particular host is completed , I can power on the VM.

The same has to be repeated for the remaining 4 hosts.

Let me know if my understanding is right & the procedure is correct.

If any other recommendations please let me know.

[senthilkumarms8]

Put your DRS in manual mode, so that the EAM will not automatically reboot the host. once you set the DRS in manual mode then go ahead and click the upgrade option, as soon as the VIBs are upgraded, manually migrate the VMs from one host to another host and reboot the host manually. After the reboot check the host status and migrate back the VMs to the same host. reapte the same steps for all other hosts.

Note: non rebooted host status will show it as "reboot pending" in the host summary page. New VIBs will not taken effect until you reboot the host.

[rajeevsrikant]

Thanks.

Manual DRS will be difficult & time consuming. Since the no of VMs in each host is many it will be difficult to calculate which hosts has the available resources & then moving them back will also be difficult & time consuming.

My plan is as below.

- Will kick on the cluster host upgrade.  The DRS will kick start the upgrade for any of the hosts.

- Once the host is identified, then I will login manually to the storage VM & do the shutdown.

- Once the maintenance is completed i will power on the storage VM/

- I will repeat this for rest of all the hosts.

Let me know if this makes sense.

[senthilkumarms8]

This makes sense. But how will you make sure that the storage VMs will not be kicked out by DRS? also the EAM agent will try to reboot the hosts in parallel.

[RaymundoEC]

I believe you should give a try for your query in vSphere Forum.

[Sreec]

What is the function of storage VM ? Providing space to VM's on specific host or are they running multipathing s/w ? Is it running on local or shared datastore ?