ESXi 5 boot from SAN with 2.0(1t) on B440 M2 problem

[kkaufmach]

Hi, we have 6 ESXi 5 running B200 M2 blades running fine (with boot from san).

This week we got 2 new B440 M2 Blades. We can install ESXi 5 without problems on the desired LUN but when we reboot the server we get:

"reboot and select proper boot device or insert boot media in selected boot device and press a key"

This seems very strange since all B200 Blades have no problem with the same boot policy. We then tried several different boot orders and also tried with only one path the the storage. All fail with the message above.

We use UCS 2.0(1t).

kind regards

chris

[padramas]

Chris,

Make sure that you have configured the correct boot target WWPN ID and LUN ID in service profile boot order.

Then, under UCSM > Server > General tab > Boot order details, does Actual boot order and Configured boot order match up ?

Padma

[mipetrin]

Chris,

Also in addition to what Padma has said, ensure that all the components of the B440 have been upgraded to

2.0(1t) - in particular the adapter firmware.

Let us know how you go.

Thanks,

Michael

[ascherrer]

Hallo,
Chris and I work within the same team and I will try to answer.

padramas: yes, we confiugred the correct boot target ( WWPN ID and LUN ID are correct, we checked it more times ). When we boot from CD we see the installation LUN and can choose it for ESXi installation. The ESXi installation self, works fine.

Under UCSM->Server-->General--> Boot Order we see the configured boot order, which is our boot policy ( must be correct configured, because other 6 blades M200 M2 also use this policy and can boot from SAN without any problem).

Under actual boot order we see CD/DVD, HDD and internal EFI Shell. Under HDD there is no LUN visible.

We also do not have any flogi error. Zoning on MDS Swithces are as same as for all other blades.

michael: yes, we have upgraded both blades to 2.0(1t) version. Under Installed Firmware we have following versions installed: Adatper1-->2.0(1t), Adapter2--> 2.0(1t), BIOS-->B440.2.0.1d.0.11162011434(we also tried with other 2 BIOS versions, which are listed under Host Firmware Packagers), Board Controller-->B440100C-B4402008, CIMC Controller 2.0(1.t).

It still does not work

Any other idea?

Many thanks for your ansers,

andy

[padramas]

Hello Andy,

Actual boot order is the one used by BIOS to boot the server.

To update it, make a dummy change to configured boot order ( boot policy ) , save it, and then make changes to have the desired boot order. Now reboot the server and verify the outcome.

What CNA adapter do you have on these B440 blades ?

Padma

[ascherrer]

Hello Padma,

I created new boot policy and added it to the server. Unfortunatelly it still does not work. Is it possible to set boot from SAN in BIOS from server self ( without boot policy)?

I thought the CNA adapter role by UCS blade servers is taken by FIs

On B440 blade server I just see Adapter 1 and 2 and they have PID N20-AC0002, Serial-QCI1544A33X  and QCI1544A3KF

thanks

[ascherrer]

Adapter Name is Cisco UCS M81KR

[padramas]

Andy,

For troubleshooting purpose, let us just have single path and verify whether Cisco CNA adapter has logged into the fabric or not.

After powering on the blade, we can verify the status by

SSH into FI

connect chassisID/slotID/adapterID ( ex connect 1/1/1 )

attach-fls

vnic

login

lunmap

If it is not logged into the fabric, verify the vHBA name defined in the service profile and vHBA name in the boot policy.

http://tools.cisco.com/Support/BugToolKit/search/getBugDetails.do?method=fetchBugDetails&bugId=CSCtt22440

If it does not match, try the work around and let us know the outcome.

Padma

[ascherrer]

Padramas,

I tried all steps, which you recommended, inkl. work around for service profile and it still does not work.

I also tried with only one path, with only one HBA...and got the same result.

Another idea?

Many thanks for your help,

[padramas]

Andy,

Does it login into the fabric ?

If there are no suggestions, please open a TAC service request.

Padma

[ascherrer]

Yes, it login into the fabric.

regards,

[Robert Burns]

Andy,

If you have successfully installed the OS to the LUN, and you can see the Initiator logging into the fabric, then the path is legit.

What type of array are you using?  If you're using an Active/Passive array such as an EMC Clariion, you have to ensure the Storage Processor (SP) you're targetting in your boot policy is the "owner" of the LUN.  You can check this on the array by looking at the LUN properties.  If you're targeting B, but A is the owner, you just need to Trepass the LUN. 

The symptom you explain align with this possibility.  If this isn't the case, then I'd suggest to get a TAC engineer on webex as Padman suggested.

One other suggestion is that in your boot policy if you have "Enforce vHBA names" checked -  un-check it, re-ack the blade and attempt to boot.

Regards,

Robert

[varrow_jgillow]

Have you tried to reacknowledge the B440 M2 blades? (while they are associated to the service profile with the boot from SAN policy).  Note -- they will power cycle for that procedure.

[ascherrer]

Hello,

In next 2-3 days we are going to check whole SAN/Netapp and UCS configuration. In case of any missconfigured setting we will change it. Should it work is ok, otherwise a TAC service request will be opened.

Many thanks for your help and typs

p.s I will post the end result here

[ascherrer]

Hallo,

We could solve our boot from SAN issue. The problem was wrong zoning configuration on MDS switch

Many thanks for your help,

best regards,