Intermittent connection from vCenter to ESXi cluster while installing ESXi Patch

[EugeneChua]

Hi Everyone,

My environment has a vCenter managing a few ESXi host through cross site. i.e. vCenter at DC1 managing 4x ESXi at DC2.

As of current, I am facing intermittent connection between my vCenter and ESXi and that renders my ESXi unable to be managed.

As the network issue is beyond my means and that my customer is pushing to roll out the latest ESXi patches to fix security vulnerability, I am trying to assess a few things before proceeding to patch my servers.

I would like to know what could be the behavior during the following situation

1) Triggered maintenance mode for one of my ESXi host in a cluster through vcenter.

Q: While the virtual machines is moving to other available host halfway and my vcenter lost connection to the whole cluster, what will happen? Will the vmotion / virtual machines be affected?

note* my cluster is configured with DRS for the vmotion but not HA.

2) Triggered remediation through update manager

Q: During this process, if the vcenter lost connection to the whole cluster, what will happen?

Any response will be of help thank you!

[DCasota]

Assuming the environment is on vSphere 6.7, have a look to vSphere Health. It works to identify and resolve potential issues before they have an impact.

vCenter to ESXi over WAN connectivity has some limitations and therefore it is recommended to deploy VCSA at the same location where the ESXi hosts are deployed.

By studying network traffic (have a look to ports.vmware.com), consider to use port groups to properly isolate network traffic. Have a look to the vmkernel settings on the ESXi hosts and which traffic uses which vmkernel port. Use vmkping (see VMware Knowledge Base) to test network connectivity.

Considering the vCenter behaviour for udp heartbeats, see ESXi host disconnects intermittently from vCenter Server (1005757) | VMware KB .

[IRIX201110141 · Accepted solution]

For patching ROBO Hosts thats what is the "Staging" option is good for. It transfer as most patches as possible to the ESXi first and than later you can try to install them and verify if the Host comes back before starting with the next.

But even if you choose the normal way ... ESXi always try to fetch the patches from VUM first and download them. If  your WAN connections breaks the download stops and the job timeout.

About DRS... it always needs a vCenter to start a vMotion. Question is what happend during a vMotion if vCenter connection went away about your flaky WAN. I dont now what will happend. I assume that once the Jobs start the vcenter isnt needed any more.

About HA. vCenter is only needed for setup and configuring the HA. After that HA works between the Hosts and vCenter connectivity isnt needed. So it will work perfectly in your case.

Regards,
Joerg