Feature request: HyperFlex better integration with vLCM / VUM

[Riaan van Niekerk]

I know HX Connect can remediate an ESXi cluster, put hosts into HX maintenance mode 1-by-1, however:

a) patching our HX environments is very different from the rest of our VMware estate (and runs the risk of a less experienced administrator putting a host in regular maintenance mode)

b) Not all ESXi patches receive Cisco offline bundles for use in HX Connect.

As a workaround, when using vLCM to patch, we can put hosts 1-by-1 into HX Maintenance Mode using the HX plugin and remediate them via VUM. 

Are there any plans to make HyperFlex aware of vSphere Life Cycle Manager (formerly vSphere Update Manager), e.g. that a whole cluster can be remediated using vLCM, instead of one host at a time, or is HX architecturally incompatible with vLCM in this regard?

[jhvillam]

Hello riaan.vanniekerk!

There are not plans to support it and it is not in our roadmap either, the reason behind this is that it can do more harm than good since pushing ESXi images without the right VIBs for Hyperflex to operate with ESXi can take down HX clusters

"Using VMware Update Manager (VUM) or VMware Lifecycle Manager (vLCM) for upgrading the ESXi on HyperFlex node is not supported. Using these upgrade methods may delete Cisco custom drivers and cause cluster outages. We recommend using Cisco Intersight or HyperFlex Connect for ESXi upgrades including the security patches from VMware or manually installing patches using the offline zip bundle with ESXCLI commands."

https://www.cisco.com/c/en/us/td/docs/hyperconverged_systems/HyperFlex_HX_DataPlatformSoftware/HyperFlex_upgrade_guide/4-5/b-HX-Upgrade-Guide-for-VMware-ESXi-4-5/m-upgrade-vsphere.html

[Riaan van Niekerk]

Thank you, jhvillam

My question was specific to patching / updating, not upgrading, but the same logic would apply. We currently upgrade using HX Connect and update using VUM/vLCM per-host remediation with HX connect. We will probably review our operational standards and just use HX Connect for both use cases. 

[Steven Tardy · Accepted solution]

All VMware ESXi patch .zip files (from vmware.com) can be used within HX connect to upgrade ESXi.
HX connect will manage the ESXi upgrade of all nodes in the cluster in a rolling reboot fashion.

Do you have examples of VMware ESXi patches NOT working in HX connect?

Upgrading through HX connect seems like even less work compared to upgrading via vLCM.

Overall this is more like a VMware issue/request to have VMware vLCM/VUM be "HX aware".

[Riaan van Niekerk]

Thank you Steven

It helps that HX Connect will work with all VMware offline bundles and takes away one major reason for wanting to use VUM/vCLI for updates. I have not tried remediating an HX cluster with a VMware provided offline bundle, but can accept that it will work, assuming the zip file contest (directory structure, manifests etc.) is the same for what VUM/vCLI and HX Connect expect. 

I guess we need to decide which of the following two processes we want to follow to patch our HX Clusters:

a) Use the same tool as used by non-HX clusters: VUM/vLCM, with the advantage of not having to manually download patches (they are already automatically downloaded) but with many caveats, e.g. (i) after putting host in vSphere MM, put host in HX MM using vSphere plugin or HX Connect (ii) can only remediate one host at a time, vs

b) Use HX Connect for updates as well (in addition to upgrades). It has the manual step of downloading and importing the offline bundle, but has the advantages of (i) vSphere & HX MM being handled for you (ii) being able to remediate the whole cluster. 

Based on weighing up the above pros and cons, and the feedback from you, I will recommend (b) to our team. It is different, but better and more fit for purpose. 

We will try updating using VMware supplied patches in HX Connect during our next patch cycle (if an equivalent Cisco custom offline bundle is not available by then).

> Overall this is more like a VMware issue/request to have VMware vLCM/VUM be "HX aware".

Not sure if this statement was serious or meant in jest / a “host take”. It would be very rare for a downstream vendor (VMware) to make adjustments to its stack to fix an issue experienced by upstream hardware or software vendors. VMware provide the APIs, frameworks and standards on top of which 3rd party vendors develop their solutions and integrate with vSphere. If there were something missing on the vSphere side, vendors could lobby VMware to expand or expose the functionality so they can expand their solutions. E.g. the inability to remediate a whole cluster using VUM/vLCM is both a HyperFlex and Nutanix limitation. For customers that will be migrating from HyperFlex to Nutanix, they need to learn another tool / process (after moving alls VM off the host, you manually power off the Nutanix CVM then put the host in vSphere MM, after which you can remediate one host at a time with VUM/vLCM).