UCS C220 M3 - Need to change RAID config from 0 to 1

[gilbreencisco]

I recently started a new position where we have three UCS C220 M3 servers running ESX 5.5.  The three are in a cluster and share datastores on a NetApp appliance.  Each server has two 600GB HDs installed.  My predecessor left a note stating that one of the servers was misconfigured with its two HDs in RAID 0 but should be in RAID 1.  I am fairly new to both UCS servers and ESX.  What is the recommended way to proceed in changing the RAID configuration?  I assume the steps include:

- Migrate the VMs that are on the machine via vCenter Server to one of the other servers

- Delete the existing RAID 0 virtual drive

- Add a new virtual drive in RAID 1 config

- Reinstall ESX

- Migrate the VMs back to the original server

I have probably simplified the process and/or left out critical steps.  Any recommended steps or direction to accomplish the task at hand?

Thanks.

[Kirk J]

If you only have 1 drive in the raid 0, you can add an additional drive, and have the LSI webbios utility migrate to a 2 drive raid 1.

As you have two drives in a raid 0, you will need to delete and recreate the VD after migrating your guestVMs to some other storage, as you have already concluded.

Your steps are correct.

You might want to remove the host from vcenter, prior to deleting and recreating raid VD, as this should release the license assigned to it after a few minutes.

Thanks,

Kirk...

[gilbreencisco]

Thanks Kirk.

I have migrated the VMs to another host, removed the host from the vCenter and am ready to remove and recreate the VHD.  When I log in to the Cisco Integrated Management Controller web admin page and go to the Storage tab, on the Virtual Drive Info, I can see the option to Delete Virtual Drive.  But I am not sure if that is where I do it.  Since the server only has the one VHD, I assume it also hold the server OS, but that is just a guess.  The RAID controller is listed as a Cisco USCS RAID SAS 2008M-8i.

I know these are basic questions, but what are the steps to delete the VHD and create the new VHD in RAID 1.  Is it done via the web interface or will it have to be done via the CLI?  If via the CLI, I am not sure of the steps.  I have done a search but have not had luck finding steps for the installed controller.

I have attached pics of the physical drive info tab and the virtual drive info tab.

Thanks.

[gilbreencisco]

I did more digging and found how to recreate the VHD in RAID 1.  I went in through the management console and went into setup.  There I was able to launch the LSI webbios and was able to look around until I saw what I needed to do.  Once it was done, I was still able to boot the server up, access the web CIMC, start the KVM console, and reinstall ESXi.  One thing that tripped me up is that the KVM console would not open with the current shipping version of Java.  I had to uninstall it, install version 7 and then was able to access the KVM.

So, everything is back to how it was, but now on RAID 1.

Gilbert

[Kirk J]

Glad to hear you got it up and running on a redundant storage solution.

Yea, Java versions above 7 may not work well unless you have the later HUU firmware applied to the server (i.e 2.08 and higher).

Thanks,

Kirk

[gilbreencisco]

Kirk,

Thanks again.  You bring up a great point.  I also noticed that the firmware we have installed is version 1.5(1f) which is very behind the current version.  I read this morning that it is recommended to update both CIMC and the BIOS firmware using the HUU iso.

What is the recommendation for upgrading the firmware in regards to ESXi?  Does the firmware affect what is loaded on the HD?  I assume to be safe, I should move the VMs off the server just in case, but don't know if I am being overly cautious.

Lastly, is there any concern in updating wholesale form 1.5(1f) to the current version 2.0(9l)?

Gilbert

[Kirk J]

Hi Gilbert.

Definitely consult the UCS interop matrix at http://www.cisco.com/web/techdoc/ucs/interoperability/matrix/matrix.html to make sure the ESXi version, the HUU/Firmware, and the applicable driver versions they require.

The HUU can update the CIMC, BIOS, raid controllers, add in NICs, harddrive firmware, etc.  We recommend letting the HUU 'upgrade all'

In regards to the HUU process for that large a version jump, the recommended process is stepped.  See a previous post I had that addressed this at: https://supportforums.cisco.com/discussion/13050796/upgrading-ucs-c220-m3-server-firmware-version-153d-209f

Thanks,

Kirk

[gilbreencisco]

Hi Kirk,

Thanks for the info and links.  I had checked the requirements for ESXi and planned on updating it once I updated the host.

I perhaps should have waited for your response as I went ahead and did the jump from 1.5(1f) to 2.0(9l) in one step.  All seems to be working fine, with the exception of booting from a virtual ISO.  I was attempting to update ESXi via an ISO, but when I reboot the server and choose F6 to get to the boot menu, I briefly see the boot menu for a split second but then it proceeds to boot from the RAID.  I was able to work around it by updating ESXi via the vSphere client update manager.

Not sure if I am not timing it right, or if is an issue resulting from the upgrade.  Is a rollback to the 1.5(4e) even possible or am I best just to leave it along at this point?

I have two more servers to upgrade, so I will certainly do the stepped upgrade procedure as outlined your other posting.

I do appreciate all the assistance.

Gilbert

[gilbreencisco]

As an update, I have upgraded the two other servers, using the stepped method.  However, they are both exhibiting the same behavior regarding the boot menu only appearing for a split second before I can make a change to it.  However, for all three machines, it is sporadic - sometimes it shows the boot menu and other times, it just briefly flashes it and continues to boot.

I have found the only way I can get the server to boot without the boot menu is by hitting F2 and accessing the Setup Menu and choosing the Boot Options tab there.  From there, I have to manually move the CIMC Mapped DVD to the top of the list, save and exit (F10) and then it will boot from the loaded ISO.

The option to select the Boot Menu appears twice - once at the beginning of the cycle and once again after all the ethernet adapters are loaded.  Should one wait until the second time before selecting F6 for the Boot Menu?

[Kirk J]

The boot menu, normally won't let you into it until all the option roms have posted, and I do think I recall that F2/F6/F8 available options popping up twice.

You might want to confirm you have disabled any unneeded Option ROMs (i.e. addin nics, LOM NICs, etc) and just enable the option rom for the PCI-E slot the raid controller is in.  Does you current boot order as defined in the CIMC, BIOS, boot order, have a local entry for local CD/DVD 1st in the order? If this is present, it should boot to your mounted ISO, without having to F6/boot menu, to select it.

Thanks,

Kirk...