Upgrading the Hyperflex Cluster
We have now completed the first step which was to upload the UCSM Infrastructure A-bundle, Blade B-bundle and Rack C-bundle for version 4.0(1a) to the Fabric Interconnects.
Next step is to check the vMotion between hosts to ensure when you start to failover you won’t have any issues. If your cluster is already in production then more than likely this will not be an issue but it’s always worth checking from each host.
Now it’s time to upgrade the UCS infrastructure.
Modifying the Firmware Packages in Cisco UCS Manager
Now it is time to modify the firmware packages so that we can manually stage the firmware and start the upgrade. We will start UCS Manager.
Log into UCS Manager
Click on Equipment, then Firmware Management, then Firmware Auto Install
Next we need to click on Install Infrastructure Firmware
The Install Infrastructure Firmware wizard will now start, review the warnings and resolve any issues it may present. I have reviewed all of my hardware versions so it’s safe to select Ignore All and click Next.
From the Infra Pack drop down box select the new version you want to install, in our case 4.0(1b). Then tick the Upgrade Now box and click Finish.
At this stage I don’t have any service packs we need apply with the base firmware so I will leave that blank. Service packs are patches that you can use to apply security updates to UCS Manager infrastructure and server components.
Then click OK
This will upgrade the UCS Manager software first so your connection will drop out at this point. Wait a few minutes before attempting to log back in. Don’t worry everything is still online as it is just upgrading the management software side.
Next it will look for any IO modules you have present in your 5108 chassis (if connected) and we need to wait while each one has the new software staged to it. They will display Pending Next Boot once complete but the new version will not be activated until the FI’s are each rebooted.
After a few minutes you should be able to log back into UCS Manager. Note the new version displayed 4.0(1b)
Next we need to wait while the subordinate fabric interconnect updates. If you click on Equipment then Fabric Interconnects, then Installed Firmware you will see that the subordinate is activating.
All traffic will now be forwarded to the primary FI while the subordinate is rebooted. After it has activated the startup version will be listed as 4.01b and the status will be set to Ready.
If you have a blade chassis present you will have to wait for the firmware to be staged to the IO modules and for the ones connected to the subordinate FI to display Ready. The ones connected to the Primary will still show Pending Next Reboot.
Wait until all errors have cleared from UCS Manager before continuing.
Next you will notice the Pending Activities alarm at the top of the UCS Manager page. Click on the Pending Activities image.
Click on Fabric Interconnects and then click Reboot Now.
Click Yes to continue
Then Ok one more time for good luck.
Can’t believe you just rebooted your primary FI.. What were you thinking? Just kidding its fine.
This will now reboot the Primary FI and the Subordinate will become the Primary. During the failover all traffic will now be directed via the new Primary FI. Don’t worry everything will still be connected!
The old Primary FI will now show Activating and will then reboot. At this point the failover kicks in your connection to UCS Manager will drop so just wait a few minutes and then log back into UCS Manager.
At this point you may notice your hosts reporting network uplink redundancy lost. Don’t worry the connectivity will be restored shortly.
Once the failover is complete you will be able to log back into UCS Manager. Click on Equipment then Firmware Management then Installed Firmware. Expand out the Fabric Interconnects.
Check that the Subordinate has now become the primary. Then wait for the activation status of the other FI to change to Ready. Wait for any IO Module activation’s to complete.
Once this is done the error counts at the top of UCS Manager should all go back to 0. Which is what all UCS admins love to see whenever they log into UCS.
To confirm that everything is now back online and healthy switch to your vSphere Web Client. Click On Home, then vCenter Inventory Lists, then Cisco Hyperflex Systems, then Cisco HX Data Platform, then Cluster and then Summary. This should now show the Cluster Status is Online and the Health Status is Healthy.
The HX Data Platform Installer and the Server Upgrades will be detailed in Part 3