UCS B200M3 and Flex Flash (SDCard)

[previfrance]

Hello,

I have migrate fews UCS B200M3 to boot from SAN to SD Card since it is supported by firmware 2.1.

I'm running the last firmware, 2.1(1a) and ESXi 5.1.0, 1157734 with Cisco Custom image.

After few days, my ESX lost connectivity with SD Card and in vSphere it says : Lost connectivity to the device mpx.vmhba32:C0:T0:L0 backing the boot filesystem ... (see attachement) (I have the error an all my ESX on SD Card).

If I reboot ESX ou restart management agent, the message is gone and all is working.

There is a problem with ESXi 5.1 on SDCard and B200M3 ?

Thank You

[smcquerr]

Vasim,

Thank you so much for your response and the additional information.  I am troubled that you were asked for business reason and I will investigate why this was done.  Only enhancement requests are evaluated against the expected percentage of customer use, customer impact, and/or a business case.  A defect that prevents the expected operation of our product will always be corrected as soon as possible and integrated into the next possible release train as detailed in my previous response.  Bug ID CSCuh33982 was identified as severe and was fixed on 9/27/13.  Bug ID CSCui77602 was identified to be the same as CSCuh33982 and resolved on 9/16/13 the priority should have been marked equivalent.

The next release for standalone release was 1.5(3d) released on 10/14/13  and the next available release for UCSM will be version 2.2 which is currently scheduled for a December 2013 release (subject to change).  The preivous UCSM relese was 2.1(3d) which ws 10/10/13 so there was no way the fix could have been pulled in earlier even though the fix was completed a month earlier due to the Q/A and release schedule.  Again I am sorry that you were misinformed and I stand corrected in my earlier assertion that the BU would not have asked for a business case.  I am truly amazed that it happened and I will look into what you were told in greater detail.  Once again I assure you that it is our intention to correct and incorporate bug fixes as quickly as possible.

Thank you,

Steve McQuerry

UCS - Technical Marketing

[Keny Perez · Accepted solution]

Loic and everybody else following this thread,

The bug CSCui77602 was finally marked as a duplicate of  CSCui82679 , which is now marked as fixed.

UCSM 2.2 was released yesterday:

http://software.cisco.com/download/release.html?mdfid=283612660&softwareid=283655658&release=2.2

Here you have the release notes of UCSM 2.2 :

http://www.cisco.com/en/US/docs/unified_computing/ucs/release/notes/ucs_2_2_rn.html

And even when I do not see the bug mentioned as part of the resolved issues, I will internally make sure that is corrected.  Internally the bug says it is fixed and where, same way it was fixed for the rack servers in CSCuh33982.

Please remember that on C series, the issue is resolved with the upgrade of CIMC AND the reset of the FlexFlash, so this might also be needed on B series if you see the patch is applied and the issue continues.  I say "This might" cause the new major released was just made available yesterday but the fix is there on C series and was a definite solution.

We appreciate your patience with this issue and would appreciate any feedback after the upgrade is applied to your domains, while in the mean time apologize for the problems this might have caused so far.

-Kenny

[previfrance]

Thanks Keny,

I'll try ASAP and I'll give you my feedback in a few days !

[previfrance]

Keny,

How can I reset the flex flash on UCS B200M3 ?

Thanks

[Keny Perez · Accepted solution]

Loic,

Here is the process that needs to be performed to guarantee the fix.

The SD controller firmware needs to run the version 1.2 build 257-1 (This can be found on the server’s inventory storage tab.).  The SD controller firmware is part of CIMC/BMC image, and is programmed in to the SD Controller on SD Controller reset.

If the “running” CIMC/BMC version is not 2.2(1) when the board update is issued the SD Controller will not be update to the new version with the CSCui77602 and CSCui82679 fixes.

<< Click on the image.

If the version 1.2 build 257-1 is not running do one of the following:

Option 1:

•1)      Verify CIMC/BMC is running 2.2(1)

•2)       Do a slot reset or reinsert the blade (The loss of standby power causes SD Controller reset)

Option 2:

•1)      Verify CIMC/BMC is running 2.2(1)

•2)      Re-issue the board update on the server.

Both options will cause the board to be powered off but we hope to improved this process soon.

Special thanks to Kevin Babst for the information provided.

-Kenny

[previfrance]

Thank you again Keny,

After re insert the blade, the running version is 1.2 build 257-2 (not 257-1 like you said ?) (before it was 1.2 build 254.2).

I'm waiting a couple of days to see if the bug is realy resolved !

Thanks

[John Blackmore]

I initially had trouble updating to 2.2(1b) and recieved Error F0327 "service profile configuration failed due to insufficent-resources.

If it helps here is the procedure I did to get the servers already in pools to take the firmware.

To update the firmware individually update the firmware on a particular server with the following steps after trying to apply firmware:

1. I put into maintenance mode and shutdown the server
2. Noting the chassis/blade I disassociated service profile
3. Delete server from server pool
4. Decommission server then resolve slot issue by clicking on the "here" link (this is where the firmware is updating)
5. Add server back to server pool (make sure FSM steps are all complete on previous step)
6. Re-associate with service profile
7. Verify server is booting up (click boot server on kvm if needed) and check firmware

My flexflashed ended up at and all my other hardware looks updated correctly.

Kenny, was there a better way to fix the error or do you see any issues with how I got 2.2 shoe horned in?

[Keny Perez]

John,

Why did you have to remove the server from the pool of servers? Where you having any particular error message ?

The best and easiest way to have this updated are the 2 options I pasted above, for what I know this process will be improved in the future to make it easier.

-Kenny

[previfrance]

Bad news for me ! After update my fabric's interconnect and IOM module with the last firmware, a few hours after, all my ESX in mi second chassis was unable to connect to vCenter, it looks like they cannot connect to the SAN and PING was ok but all VM does"nt work.

After rebooting ESXs, they hang on "USB Arbritator" for 3/5 minutes and my problerm is still present (cannot connect to vcenter and cannot run any VM).

To fix my problem, I had to rollback to 2.1 only for fabric interconnect and IOM Module and all is ok.

Edit : I think this is related to default zonning, I'll investigate

Default Zoning is Not Supported in Cisco UCS, Release 2.1(1a) and Later Release

[Keny Perez]

Yeah Loic, if that is the issue Robert Burns gave the heads up here 

https://supportforums.cisco.com/thread/2185462

-Kenny

[previfrance]

I was wrong, I don't use default zoning and my UCS use firmware 2.1(2a) so default vlan is already disabled.

I'll open a Cisco TAC to because I cannot upgrade Fabric Interconnects.

[Keny Perez]

Loic,

The key is the version 1.2 build 257-x, older versions will show build 255 and 254.

-Kenny

[previfrance]

Keny,

I can't update "IO Module" with 2.2(1b) firmware.

When I try to update I have this error :

John, do you have updated the IO Module ?

Thanks

[John Blackmore]

This is what mine looks like, to the best I can tell its all updated to 2.2(1b).  I did start with the Equipment->firmware manabement->firmware auto install and did "install infrastrucure management" first before I did anything with the blades firmware.

Hope this helps.

[previfrance]

Thanks John, It Works !