VxRail Cluster Shutdown occurs after PowerChute has completed the shutdown of User VMs. This is initiated by the NMC using the cluster shutdown REST API call via VxRail Manager.

VxRail Shutdown and Startup
The Cluster Shutdown Delay is the delay used by the NMC to allow sufficient time for the PowerChute VM to shut down, before the NMC issues a shutdown command to the VxRail Manager. The default value is 120 seconds.
|
|
The Cluster Shutdown Delay should be set to 0 seconds for VxRail stretched cluster configurations. |
|
When a shutdown command file and/or SSH action is configured in a standard cluster configuration, the Cluster Shutdown Duration provided must be sufficient enough to allow the command file/SSH action to run and the PowerChute VM to shut down. If a sufficient Duration is not provided, the VxRail cluster will not shut down successfully. |
|
Using the Cluster Shutdown Duration field, you must allow sufficient time for your Cluster to gracefully shut down once the VxRail Manager has been requested to stop the Cluster. It is recommended that you manually test the time needed to stop your Cluster, and specify that as the Shutdown Duration. The default value is 300 seconds. In a VxRail stretched cluster configuration, PowerChute waits for the Cluster Shutdown Duration to elapse before proceeding with the next step of the shutdown sequence. If the duration provided is not sufficient, cluster shutdown may be unsuccessful. |
Depending on your configuration, PowerChute is restarted after a shutdown by one of two methods. If the PowerChute VM and vCenter Server VM are deployed inside the VxRail cluster, PowerChute restarts via a scheduled task after vCenter Server starts. For all other configurations, PowerChute is restarted via VMware's auto-start policy.
|
|
NOTE: When the Dell VxRail cluster is restarted, the vSAN datastore is unavailable until all the cluster services are running. In a stretched cluster configuration, it is highly recommended that a value is set for delay_before_vmstartup in the [HostConfigSettings] section of the PowerChute configuration file. This setting ensures that PowerChute restarts User VMs when the vSAN datastore is accessible. |
Using this method, PowerChute registers a scheduled task during shutdown to start the PowerChute VM after vCenter Server with a given delay. The delay can be configured in the PowerChute configuration file (pcnsconfig.ini) using the "scheduled_start_delay" setting in the VxRailCluster section.
Once set, the scheduled task can be viewed in vCenter by selecting the PowerChute VM, and navigating to Configure > Scheduled Tasks.
If the VxRail cluster services have not fully started, this may be unsuccessful and the PowerChute VM will need to be started manually.
Using this method, PowerChute registers itself with the Virtual Machine Startup and Shutdown policies on vCenter and the PowerChute VM is restarted with its host. The Virtual Machine Startup and Shutdown policies can be viewed in vCenter by selecting a cluster host, clicking Configure, and navigating to Virtual Machines > VM Startup/Shutdown.
|
|
The auto-start policy will be enabled if PowerChute is deployed as a VM on any VMware setup. |
|
|
Ensure that the auto-start feature is enabled in ESXi. If it is not enabled in ESXi, PowerChute may not successfully restart after a shutdown. To enable auto-start in ESXi, click on a cluster host, click System, click Autostart, and set Enabled to "Yes". NOTE: This feature cannot be enabled when HA is enabled. |
VxRail manages restarting the VxRail cluster and system VMs. VMware auto-start will attempt to power on the PowerChute VM when it’s host reboots. When PowerChute is deployed inside the VxRail cluser this will not work as the hosts will be in maintenance mode after they are powered on. For details on how to have PowerChute automatically power on in this scenario, see Knowledge base article FAQ000258432.
The startup delay has a default value of 0 seconds. This can be configured in the PowerChute configuration file (pcnsconfig.ini) using the "autostart_delay" setting in the VxRailCluster section.
|
|
When Dell VxRail and Management hosts are powered on at the same time in a stretched cluster configuration, the datastore may become inaccessible and cluster startup may not work as expected. To prevent this issue, ensure that the Management host is started first followed by the VxRail hosts so the Witness host and vCenter Server are available when the VxRail hosts start up. For more information on how this can be achieved, see the "Cannot start Cluster" section in the Dell VxRail Troubleshooting topic. |