Solved

First time using vmware upgrade manager on cluster

  • 13 October 2015
  • 3 replies
  • 1196 views

Badge +9
Just had the cluster installed. I'm going to patch my 5.5 installs, I put the first node into maintance mode and its just waiting for the CVM to power off or move. Should the CVM shutdown on its own or do I need to power it off before I patch and reboot?

thanks,jb
icon

Best answer by tduval 14 October 2015, 16:07

View original

This topic has been closed for comments

3 replies

Userlevel 1
Badge +10
Hello , as explained in this post http://next.nutanix.com/t5/Product-Ideas/CVM-as-Agent-VM/m-p/199 you have to properly shutdown the CVM manually to perform maintenance mode on a node.

As I'm running my resources on Hyper-V cluster, I take advantage of your post to ask if there's a way to tell SCVMM to not put CVM into saved states during maintenance mode but instead to shutdown/power off it ?
Badge +9
H iguys,

I don't know if you finally had your answer.

I recommand using Nutanix documentation materials before doing any kind of maintenance on OS part of your nutablock !

You have some command to gracefully upgrade your cluster. You have the one click upgrade too ! CVM / hypervisor / acropolis base are tightly coupled so upgrading any part should be done respecting the block and procedures validated by well nutanix kids on the block ;)

seriously IMO, don't do if you're not sure : on vsphere administration guide for acropolis 4.5
  • p24 = Patching ESXi hosts in a cluster.
  • p29 = vSphere Upgrade.
I bet you have the same for Hyper-V

enjoy !

thomas
Userlevel 2
Badge +7
One thing I did find with the 1 click upgrade on my clusters to 5.5u2 was that if you should have a vm that cannot move for some reason the 1 click upgrade will fail with no notification.

We had some vm's that had virtual nics tied to no physicals and the vswitch only on one host.

I have discussed this with our SE and they are going to look into putting a pre-check or a wait and notify if it runs into one of these sessions instead of failing.

But on clusters where this wasn't an issue the 1 click worked perfectly. I moved all the sessiosn and then shutdown the CVM. and It waited until the the last cvm was back up and running and part of the cluster until starting the upgrade on the next host.