VxRail: How to Properly Shut Down a VxRail Cluster Versions 7.0 U3 to 8.0

Summary: This article provides steps to avoid issues and the procedure to shut down and power-on a VxRail cluster. This is for VxRail clusters starting from version 7.0 U3 (using vSAN plug-in). The procedure works with an internal or external Domain Name System (DNS) and an internal or external vCenter. ...

This article applies to This article does not apply to This article is not tied to any specific product. Not all product versions are identified in this article.

Check out other resources

Instructions

Note: If DNS is hosted on VxRail, shutdown of DNS should be performed manually before shutting down the cluster. If DNS is hosted solely on the VxRail cluster follow VxRail: How to correctly power off a VxRail cluster when the VxRail plugin is not working

Procedure to shut down:

Find the vCenter virtual machine (VM) to see on which node the VM is running (this is needed for later).
Here is a figure showing the vCenter VM - running node. In this example, the vCenter is running on node X.
Disable High Availability (HA):
Here is a figure vSphere HA configuration - disabled.
Find the Cluster reference:
In this example, it is: domain-c9
Do a retreat mode to remove the vSphere Cluster Services (VCLS) VMs:
The figure below showing the retreat mode configuration and options:
True or False is case-sensitive, do not use (false or true), use with the first letter uppercase.
If your vCenter does not have the parameter, add it here.
```
Name: config.vcls.clusters.domain-c9.enabled  Value: True/False
```
This adds the parameter to the cluster name: domain-c9 and use the value as False to remove the VCLS VMs. If running VxRail version 8.x, the retreat mode is easier: Select the cluster, configure, vSphere cluster services, general, VCLS mode, edit VCLS mode
Check if the VMs are gone:

The retreat mode should shut down the VMs and remove it. It can take a few minutes.
Check the nodes iDRAC Ip-address, this is needed this for later:
Shut down all the virtual management system (VMS) inside the cluster, including VxRail Manager, except for the vCenter VM.
Shut down the cluster using the vSAN plug-in:

The process can take 10-40 minutes. Check the process by accessing the iDRAC and checking if the node is still online:

Procedure for Power-On:

Power on the nodes again using the iDRAC:
Monitor the iDRAC console until all nodes are finished booting.
Access the node where the vCenter, DNS, and VxRM were running and check if vSAN is up already or not (If your DNS is external, check if it is running):

In this case, the vSAN was not ready yet and it shows all the VMS as invalid. This process can take time 10-30 minutes. If doing a refresh after 30 minutes in your browser still shows the same behavior, open a VxRail support ticket for assistance. After vSAN is loaded, the VMs will show as the following example:
Power on vCenter, DNS, and VxRail Manager VMS vCenter can take 5-20 minutes to fully start. Log-in into vCenter:
Access vCenter:
Restart vSAN services:
Enable HA again:
Do the retreat mode again to setup the VCLS VMs:

True or False is case-sensitive, do not use (false or true), ensure to use it with the first letter as uppercase.
Put True to deploy the VCLS VMs again, and check if the VMS is there, this process can take a few minutes: Check VCLS VMS:
If the VxRM is already running, check the plug-in. If it is not working yet, power it on. VxRM can take 5-20 minutes to fully load:
Check the vSAN skyline health: