This article is not technically focused, I wanted to cover in here some of the common failures I see at customers regarding VCF upgrade failures.
Pre-Check
The VCF pre-check is an excellent tool to perform some basic checks on the environment before the upgrade progresses. The pre-check will also stop the upgrade if there is an issue identified that that could put the upgrade at risk. These are often things like vSAN firmware mismatch etc.
Do not ignore the pre-check and although it lets you skip some items it identifies. My advice is to always get a clean pre-check before the upgrade starts.
HP Synergy Customers
HPe often release SPP upgrades to upgrade the firmware of the Synergy platform, that tie in with major VCF releases. This is not always the case but has been for the last few 3.* upgrades. Check with HPe the order for the SPP to be applied and if there are any known issues that an SPP upgrade will resolve.
Vendor/Custom ESXi ISO images
Check with your vendor for the latest ISO release from there website. VMware also list the vendor ISO images on your “My VMware” login. Vendors appear to be approaching this in two ways based on their partnership with VMware or their own resource limitations.
- Major release:
- Some vendors only release a Vendor ISO for major ESXi version uplifts. This could be an ESXi 6.7U1, U2 and U3 release cycle.
- Some vendors may only release a base version of the release and not for the update packages.
- EP (express patch versions)
- Some vendors keep with the VMware releases and will release ISO images with the same EP updates as the VMware provided images.
- Customer build (user) images
- Vendors may use the VMware ESXi image but require the addition of drivers and/or VIBS.
Check with your vendor what image is needed, a VMware ISO or one of the above vendor ISO methods.
Once the correct image option has been selected from the above, the ISO needs to be imported in to VCF. There are instructions on the VMware site located here. https://docs.vmware.com/en/VMware-Cloud-Foundation/3.10/vcf-310-upgrade/GUID-B639896D-B4F0-4758-A02B-AA94FA6FEF1F.html
Common questions around custom ISO images
Can I use a major release vendor ISO and patch using VUM?
The quick answer is no…. The ISO image will need to be injected into SDDC manager using a specific build number for the update you are applying. Otherwise it won’t use the image. VUM is also technically not to be used in VCF.
How do I add updates to the Vendor ISO?
This process hasn’t changed for a long time and the process is still to use VMware image builder https://docs.vmware.com/en/VMware-vSphere/6.5/com.vmware.vsphere.install.doc/GUID-48AC6D6A-B936-4585-8720-A1F344E366F9.html
Does the image build number matter?
Again quick answer is yes…..the SDDC manager looks at the build number provides in the upgrade BOM and will look for the build number.
My custom image has a different build number than the BOM
The process linked above, to import the custom ISO has an option to link the custom image to the build number needed for the VCF BOM. This is then recognized by the SDDC update manager and used in the upgrade.
Check the VCF version release notes
The upgrade process for VCF will upgrade several components in the VCF environment. However not every update will upgrade all the components. This can result in some upgrades taking less than 30 mins and some taking much longer.
It is advised to look at the release notes on the VMware docs site for the version you are moving from, say VCF 3.9.1, and the version you are moving to. In this example VCF 3.10. Compare the product versions in the BOM and you will see the products being upgraded.
Can I control when each component is being upgraded?
No, VMware have tested the upgrade and validated the interoperability of the components in the upgrade to ensure the upgrade is successful and that nothing is upgraded in an unsupported way.
You will however have the ability to apply the updates one at a time. SDDC manager will provide the upgrade packages one after the other and so some customer pause the upgrade at these points based on change windows etc.
During the NSX upgrade can I “pause” the upgrade and control when the outage for the ESG’s occurs?
We have had several customers ask us if they can control the NSX upgrade and pause the upgrade when it gets to the ESG. There are several reasons for this. The most popular being that customers need to understand when the outage occurs so that they can manage this with the application owners. Yes, the outage is only supposed to be 15 seconds (max) but this can disrupt applications in many of our customers environments.
The way the upgrade is automated at the moment does not allow this. However, this is a feature that VMware are being pushed hard to introduce into a future build.