Boot from SAN Primary and Secondary

[Big Vern]

We running cisco b200 M$ blades connected to 6248 and cisco MDS.

The MDS is connected to an IBM V7000 SAN (4 paths to each fabric for 8 in total)

Can anyone please confirm that UCS ESXi 6.5 u1 cisco oem image should boot from either of the 2 SAN secondary targets (primary and secondary) if it cant see either of the SAN primary targets?

Ill start simple as it might get complicated, basically we had open zoning so everything could see everything and SAN boot was fine, but we then changed it to single iniaitor zones and activated it (at that point it continued to work), but on rebooting SOME of the ESXi servers could not boot. IF the answers is yes they should boot from secondary then Ill expand (dont want to pollute with too much info at first)

Asking because we have double checked the zoning and everything looks fine for the SAN secondary targets. (We cant use either of the SAN primary targets, which ill get too later)

Also we had a quick read on smart zoning - is this available for UCS/MDS and how do we configure it..

many thanks, we are new to this so be as verbose as you want/can please!

[Walter Dey]

Hi

Yes, in general this should work, meaning you have boot from SAN redundancy.

Q. Is your FI-A connected to MDS-A and FI-B to MDS-B (no cross connects A -> B, B - > A)

Q. do you have separate VSAN's (not eq. 1) for fabric A resp B

Q. How many paths do you see, when the ESXi is booted and up and running ? 4 ?

Q. I assume you're using multpath software, which one ?

[Qiese Dides · Accepted solution]

Hi Big Vern,

This is a problem I see a lot when customers switch policies. I understand you are new to this so I will try to explain everything in details - if you do end up opening a TAC case let me know and I will do my best to try to assist.

I have created the below guide in a step by step fashion to help assist in troubleshooting Boot From San Issues:

https://supportforums.cisco.com/t5/data-center-documents/troubleshooting-boot-from-san-installation/ta-p/3155048

Please go through this guide and let me know if any step does not make sense. This will tell us EXACTLY why the VHBA initiators are unable to reach the storage array.

Quick Over-view on the guide:

1) Will check if boot policy in UCS is correct

2) Will Power Cycle Server and through KVM when it is loading press F6 to go into Boot Menu

3) SSH into UCS Manager and connect to the specific chassis / blade / adapter with the issue

4) Run Lunlist command and provide output (In the guide I have all possible fail scenarios).

If you are unable to complete any of the steps in the guide let me know where you get stuck at. Once you get then Lunlist and you are still un-sure go ahead and post the output along with your boot policy output and I can get this solved for you in no time.

I hope this helps, please mark helpful solutions as correct so other members can find them easily.

- Qiese Dides

[Big Vern]

Great article thanks, I opened a TAC call before your response and that's basically what we went through and solved the issue!.

As a complete novice I did find similar articles but wanted the confirmation via TAC that I would be 100% correct when making target changes.

In our case the targets were all legitimate in the sense they existed, but we they were not zoned in on the MDS. (we had an IBM v7000 with 8 paths, we did manual RR load balancing when zoning the hosts so each host could only see 4 paths) Once we added the correct targets it all worked.

[Qiese Dides]

That would do it, I'm glad everything got resolved and it is all working now.

Let us know if you have any other questions :). Have a good week.

- Qiese Dides