This video shows how to replace a GPU within VxRail. Before you begin, ensure that your virtual machines are migrated to another host.
If your clusters are running on a standard license, then you need to manually migrate the VMs to another host. To migrate, log into vSphere Client.
Click the menu and select "Inventory". Expand the vCenter Server. Then click "VxRail-Datacenter". Now click the "VxRail-Virtual-SAN-Cluster".
And select the ESXi host that is running. If there are multiple VMs running on the ESXi host, then select the faulty host on the left pane. In the right pane, click the "VMs" tab and then the "Virtual Machines" tab. Right-click the running VM and select "Migrate".
Select the option "Change both compute resource and storage" and click "Next". Then select the compute resource as a target host and click "Next".
Now select destination storage for the migration and click "Next". Select a destination network and click "Next". Then select the applicable vMotion priority and click "Next". And finally, click "Finish".
Ensure that all the VMs are migrated. Next, you need to put the affected host in Maintenance Mode. To do this, select the "VxRail-Virtual-SAN-Cluster" from the left pane.
In the right pane, select the "Hosts" tab. And then right-click on the host that you need to shut down. Now, click "Maintenance Mode" and select "Enter Maintenance Mode".
Then click "OK" and acknowledge the "Warning". Next, you need to ensure that the cluster's health monitoring is set to enabled. To ensure this, select the "VxRail-Virtual-SAN-Cluster", click the "Configure" tab and scroll down the pane. Under "VxRail", select "Health Monitoring".
In the right pane, under "VxRail Cluster Health Monitoring", Ensure that the status is "Enabled". If not, use the slide button to enable it.
Next, you should shut down the faulty host from the vCenter Server. To do this, select the "VxRail-Virtual-SAN-Cluster". Click the "Monitor" tab and scroll down.
Under "VxRail", select "Physical View". From the right pane, select the affected host and click "Actions". From the list, select "Shutdown Host".
In the "Shutdown Host" wizard, confirm VM migration and click "Next". Then wait for the validation to complete and click "Next". Now confirm shutdown and click "Next".
The shutdown process has started. And once it is complete, click "Finish". Before you remove the component, power off the affected host and disconnect all attached peripherals.
Ensure that you always use ESD protection while working with the system. On the system cover, rotate the latch to the unlocked position. Lift the latch and remove the cover.
Hold the touch points and lift the expansion card riser from the riser connector on the system board. Unlock the left latch on the expansion card riser.
Pull the PCIe card holder latch out of the slot. Hold the expansion card by its edges and pull the card until the card edge connector disengages from the expansion card connector on the riser.
Now, hold the new card by its edges and align the card edge connector with the expansion card connector on the riser. Insert the card edge connector firmly into the expansion card connector until the card is fully seated.
Lock the left latch on the expansion card riser and push the PCIe card holder latch. Holding the touch points, align the expansion card riser with the connector on the system board.
Lower the expansion card riser into its place until it is firmly seated with the connector. Align the slots on the system cover with the tabs on the chassis.
Lower the latch to engage the cover onto the system. Then rotate the latch to the locked position. Reconnect all the cables and power up the host.
Once a GPU is replaced, you must exit Maintenance Mode. To exit, select the "VxRail-Virtual-SAN-Cluster" from the left pane. In the right pane, select the "Hosts" tab.
And right-click on the host that was previously shut down. Now click "Maintenance Mode" and then select "Exit Maintenance Mode". Once again, ensure that the cluster health monitoring is enabled.
To ensure, select the "VxRail-Virtual-SAN-Cluster". Click the "Configure" tab and scroll down the pane. Under "VxRail", select "Health Monitoring".
In the right pane, under "VxRail Cluster Health Monitoring", ensure that the status is "Enabled". If not, use the slide button to enable it. Finally, migrate the VMs back to the host.
To do this, refer to the previous section of the video where we explain the step-by-step instructions for migrating the VMs.
For more information or assistance, visit Dell.com/VxRail.
