ESXi5.5 Upgrade and patch installation - Supported-QA vertified

  • 7 October 2016
  • 5 replies
  • 1495 views

Badge +3
Hi all,

We have a 3x node Dell XC630-10 Cluster running on AOS4.5.0.2 - Starter License. These nodes are running a Dell Customised 5.5U2 VMware image. I want to upgrade these nodes to a Dell 5.5U3-A04 image build 3248547. This will have to be done manually as we do not have DRS running (Standard VMware license).

On top of the image I want to patch them to "Patch 8" level - release date - 04/AUG/2016 (Patch 9 available now). However reading the Hypervisor FAQ it seems that major/minor vSphere releases are supported and QA certified within 90 days of GA.

Therefore my question is:
After applying the Dell 5.5U3 Custom image then, in theory, I can only apply Express Patch 10 build 3568722- 22/Feb/2016?

Comments are much appreciated.

This topic has been closed for comments

5 replies

Userlevel 6
Badge +29
When you're patching ESXi, you dont apply the base images you use to install the system (as a general statement for all ESXi systems out there, including Nutanix).

Instead, there are patch bundles that will bring you up to whatever build you're aiming for.

So, you dont need to apply some Dell image, then some patch. You'd just apply the patch you want to apply and thats that.

Now, since you dont have DRS, you won't be able to patch through Prism and have it automated; however, you can apply them manually one at a time.

As far as patch certification goes, yea, the FAQ rules are there to keep us to a schedule of a reasonable amount of time to run QA against specific patches; however, you're welcome to apply patches at your pleasure, and both Dell/Nutanix as well as VMW will support you.
Badge +3
Hi Jon,

Thanks for replying to the thread.

I was pretty sure that applying the latest Dell image would bring the lastest NIC drivers, for example.

I will look into this.

Thanks again
Userlevel 6
Badge +29
I know what you're aiming for; however, that's not how ESXi upgrades really work, unfortunately. The base image when you first install it is what it is, then all upgrades go through the same patch/bundle process, unless you go and reinstall things.

Keep in mind that Nutanix, like most HCIA's, is an appliance based model, whether that is NX/SX from Nutanix, XC from Dell, or HX from Lenovo.

Meaning, you certainly can upgrade both Nutanix AOS as well as ESXi code, but the sub components like firmware and drivers may or maynot change at the same clip.

Luckily, we've planned for this, and added checks to NCC to look for out of date firmware or drivers and let you know when they need to be updated. This is why keeping NCC up to date, regardless of AOS/NOS version, is always a good idea.

Also, another thing to keep in mind is that ESXi patches will sometimes come with driver updates themselves, so you may get some of those things by "osmosis" so to speak.
Badge +3
Hi Jon,

Thanks for the replies.

Looking at the docs, in my case no-DRS the manual process is:

- Log on to one of the CVM and run #cluster status
- Disable email alerts
- HA: Enable ---> Host monitoring
- HA: Enable ---> Disable: Allow VM power on....
- Power mgmt ---> OFF
- vCentre Advanced settings: ADD: config.migrate.test.CompatibleNetworks.VMOnVirtualIntranet value to false (what this settings actually do? Is this going to have an impact on other non Nutanix hosts/clusters managed by vCentre???)

Follow the Preparing to Upgrade vSphere on a Host article

Then use VUM to pacth the host.

Repeat the process for the other 2x hosts in the cluster

Does this sound right?
Userlevel 6
Badge +29
I see the guide you've posted is for NOS 4.5

It would be worth while to make sure you're squarely up to date with the latest NOS (currently 4.7.2) and latest NCC (currently 2.3.0) before doing any of the host work. After that, do a ncc health_checks run_all to double check status before making any changes.

Also, I noticed your cluster isn't calling home with Pulse, so I'd recommend you check out your SMTP configuration in Prism, as well as the Alert Email configuration and Pulse configuration. NCC will also do some config checking in these areas too.

RE Patch process
Almost, but really you dont need to mess with vCenter. Here's my quick off the top of my head list

vMotion all your VM's to another host
log on to a single CVM
do a cvm_shutdown -p (this gracefully shuts down the CVM and related services)
One CVM is down, put host into maint mode
patch with VUM
once done, take out of maint mode, power CVM back on, and wait for services to come back online
Check services with cluster status and make sure all are up
run ncc health_checks run_all between patches if you really want to be extra paranoid safe
Proceed to the next node