UCS – Rebuild a Failed FlexFlash Array


When running a RAID1 array with two SD cards on a UCS FlexFlash controller you may encounter a situation where one of the SD cards has failed and you need to replace and rebuild the RAID array. Unfortunately, the steps to resolve this issue are slightly different then how you would with an HDD RAID array.


  • Verify the RAID array is unhealthy.
  • In UCS Manager click the Servers section and then click on the Server in question.
  • Click Inventory -> Storage
  • Click the FlexFlash controller in the list of controllers
  • Expand the FlexFlash Cards section
  • Verify the status of the RAID array and note which card has failed.









  • Physically replace the SD card
  • Go back into Servers -> Server -> Inventory -> Storage and click on the FlexFlash controller
  • Click Configure Auto-sync
  • Ensure that mirror mode is set to pair
  • Admin slot number should be set to the healthy SD card
    • If you get this step backwards you will erase your healthy SD card. The example below shows a scenario where SD card is unhealthy and the data should be mirror from 1 to 2.












  • Click Ok
  • Refresh the page and notice that the unhealthy card’s status is now Initializing
  • This will take 30-60 minutes and then the RAID array will be healthy.


UCS – Change the Bootable LUN


I recently ran into an issue in UCS Manager where I had multiple local LUNs configured and it was setting the wrong LUN as bootable. I finally ended up opening a support case as I didn’t see anywhere to change the bootable LUN.

Change Bootable LUN

The bootable LUN is controlled by the boot policy. The steps below walk you through finding the name of your local LUN and setting the boot policy to boot from that LUN.

  • Login to UCS Manager
  • Browse to the Storage Section and look at the Storage Profile
  • Within the Storage Profile you should have a list of Local LUNs. In the example below I have 1 Local LUN named OS.

Read More

Cisco HyperFlex – Change NTP Server After Deployment



Recently I was in the process of deploying a HyperFlex cluster and one of the NTP servers came back as invalid due to a missing firewall rule on the network. I wanted to proceed as I still had one valid NTP server but I wasn’t sure how difficult it would be to reconfigure NTP after the HyperFlex installation. Luckily it’s pretty easy and this page can walk you through the process.

Read More

UCS – Change Fabric Interconnect (FI) IP Addresses and/or Host Names


Recently I ran into a situation where I was required to reconfigure the Fabric Interconnects to use a different IP address and host name. Initially, I was worried that this could be fairly complex as it might be deeply ingrained in the UCS configuration. I’m happy to report that it’s one of the easiest changes I’ve ever made. I only saw it partially documented on a single blog so I wanted to create another post in case someone else is in my same situation.


  1. Serial connection to both FI-A and FI-B
  2. New IP addresses for FI-A, FI-B and the cluster
  3. A new root host name

Read More

Set-Realm UCS Errors During Firmware Upgrade


exclamation_warningHelp! I’m in the middle of upgrading the firmware on my UCS fabric interconnects and after I upgraded the first fabric interconnect I received these set-realm errors in UCS Manager:

  • set-realm-peer-failed
  • set-realm-local-failed

If you googled around you probably found a handful of completely useless Cisco articles that suggested you Google more or call support 🙁

Good news! According to Cisco support these errors can be caused by having mismatched firmware on the fabric interconnects. To verify everything is still working properly simply login with both an LDAP account and a local account. If you can, you are safe to proceed!