02-13-2017 09:32 AM - edited 03-01-2019 01:04 PM
Hello -
We have deployed service profiles (from updating template) booting from a single 120 SSD with a local disk config configuration policy set to mode "Any Configuration" and Protect Configuration is enabled.
These are UCSB-B200-M4 blades with Raid Controller UCSB-MRAID12G, running blade firmware 2.2(6e). The installed OS is vSphere 6.0.
We would like to install a second SSD into the blade(s) and mirror the disks. There are 42 blades so we ideally are looking accomplish this without needing to reinstall the OS.
Has any run into this or are there any suggestions? Many thanks in advance.
Solved! Go to Solution.
02-19-2017 05:53 AM
Ok, I think I see a potential explanation for the blade's 12Gb SAS controller's behavior.
Checking the spec sheet at http://www.cisco.com/c/dam/en/us/products/collateral/servers-unified-computing/ucs-b-series-blade-servers/b200m4-specsheet.pdf we can see there are several different 12Gb raid controller models.
The model I have been testing with is the UCSB-MRAID12G
Please note the footnote for that controller:
The UCSB-MRAID-12G RAID controller runs iMegaRAID, which has a lower-capability software stack, provides
RAID 0/1 levels, and has no support for caching. Note that you cannot upgrade from this RAID controller to the
to UCSB-MRAID12G-HE RAID controller
I don't have a UCSB-MRAID12G-HE in my lab pod to test with, but highly suspect that the UCSB-MRAID12G model isn't capable of doing the raid migration/reconstruction, being that it runs 'iMegaRaid, lower-capability software stack'.
DawnHC, can you confirm if your configuration includes the UCSB-MRAID-12G or UCSB-MRAID12G-HE model?
If it is the UCSB-MRAID-12G model, you may not have any other recourse other than using some sort of cloning/imaging tool, and re-apply the image after deleting and recreating a raid1 VD,, assuming a cloning workflow/process is actually easier than doing a reinstall.
Thanks,
Kirk
02-14-2017 07:22 AM
Greetings.
I believe this would be considered a Raid 'migration/reconfigure' in LSI/Avago terms.
While I don't cover the raid type change step, most of the steps are listed at https://supportforums.cisco.com/document/13133891/expanding-virtual-drive-12gb-sas-controller-adding-additional-physical-drives
In the steps there I didn't change the raid type, but I believe you would change from raid 0 to raid 1, and then follow the rest of the steps for adding the drive.
I'll test this in the lab post and post the steps when I get a chance.
I would try this on one of your sandbox blades first ;)
I am also assuming you have your existing single disk setup as raid 0, as opposed to jbod....
Thanks,
Kirk...
UPDATE...... I am having trouble on the lab test,,, need to try several different raid controller firmware versions. The lab attempt just stays at 0% Reconstruction.
02-14-2017 07:22 AM
Thank you, this provides some hope. Once I have permission to reboot one of these blades and get into the 'Avago MegaRAID (Cisco 12G SAS Modular Raid controller) Configuration Utility', I will take a look - and hope that it is set up as Raid 0.
Unfortunately, I am not seeing a way in UCSM to verify, only that it has a local disk configuration policy set to "Any configuration".
I did see in the UCS Raid Guide that a Raid migration is allowed from Raid 0 to Raid 1, but do appreciate running it by people in the forum.
Thanks,
Dawn
02-19-2017 05:51 AM
I have not been successful at running this on a Blade's 12Gb SAS raid controller, after trying several versions of firmware.
Using the UEFI tool (via F2) appears to launch the process, but stays stuck at 0% reconstruction.
I tested this on a UCS C240M4 rack server's 12Gb SAS controller, and it worked fine in both the UEFI tool, as well as the CIMC Storage tab ( Edit Virtual Drive) will let you migrate raid levels as well.
I will do some more checking to see what the expected functionality of the models in the blades is in regard to raid migration/reconstruction.
Thanks,
Kirk
02-19-2017 05:53 AM
Ok, I think I see a potential explanation for the blade's 12Gb SAS controller's behavior.
Checking the spec sheet at http://www.cisco.com/c/dam/en/us/products/collateral/servers-unified-computing/ucs-b-series-blade-servers/b200m4-specsheet.pdf we can see there are several different 12Gb raid controller models.
The model I have been testing with is the UCSB-MRAID12G
Please note the footnote for that controller:
The UCSB-MRAID-12G RAID controller runs iMegaRAID, which has a lower-capability software stack, provides
RAID 0/1 levels, and has no support for caching. Note that you cannot upgrade from this RAID controller to the
to UCSB-MRAID12G-HE RAID controller
I don't have a UCSB-MRAID12G-HE in my lab pod to test with, but highly suspect that the UCSB-MRAID12G model isn't capable of doing the raid migration/reconstruction, being that it runs 'iMegaRaid, lower-capability software stack'.
DawnHC, can you confirm if your configuration includes the UCSB-MRAID-12G or UCSB-MRAID12G-HE model?
If it is the UCSB-MRAID-12G model, you may not have any other recourse other than using some sort of cloning/imaging tool, and re-apply the image after deleting and recreating a raid1 VD,, assuming a cloning workflow/process is actually easier than doing a reinstall.
Thanks,
Kirk
02-20-2017 07:02 AM
Kirk -
Thank you so for much for looking into this. I can confirm that the configuration includes the UCSB-MRAID-12G model, not the HE. Also, the disk is JBOD and not RAID1. Reading that same footnote about the iMegaRAID lowered my hopes.
With a total of 66 blades between two sites, I had very much hoped for a straightforward mirroring solution to dramatically reduce the configuration time and opportunity for human error.
Thank you again,
Dawn
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide