cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements
Field Notice 70545
2668
Views
0
Helpful
13
Replies
harold.himes
Beginner

QLogic 5200 Switch

I am having problems with Boot from SAN with my QLogic 5200. TAC will not support this switch but the say it may work.

I have B200 M2 blades and VIC M81KR and FI 6120 in End-Host mode.. Has anyone worked with any thing like this?

1 ACCEPTED SOLUTION

Accepted Solutions

You are on the right track, you wont see an option to boot from san but both sites I just installed with these switches I went thru this same scenario for hours until they got it right.  I wont touch upon the boot process because you are seeing the WWN hit the fabric so that should be fine.  Then after adding the WWN to a zone or a storage group he's alllowed to joing the fabric.  At that time you need to ensure there is a LUN in the same group so he can see it, and it needs to be masked so only he sees it and sees it as LUN 0.  ESXi wants LUN 0 and it's really picky about those local drives being visible, you need to eject them to be safe.  But when the LUN is visible it will show on that screen there under remote, but just like a hard drive.  I bet you arent using the Cisco VMware image, I cannot see what version you are using but download the Cisco OEM image from VMware.  After I started using that with the vic card, my boot from san issues went away.   When your LUN is visible in the select a disk window just choose it to install to like a local disk.

So if SAN boot with ESXi and VIC, use cisco oem image of esxi, eject local hard drives, do NOT use diskless boot policy, use "anything" and eject at least until it's running, then there is a fix for esxi so local drives can exist, but if using for Cisco UC it's not supported.  Lastly it's the LUN that needs to be visible....that's it.  my money is on vmware .iso image due to drivers if your SAN array is correct..

View solution in original post

13 REPLIES 13
DAVE GENTON
Explorer

2 Customers I have visited in the past 2-3 weeks have those switches and I couldnt find anyone to support them either, but yes I have both ESXi  booting from SAN via Qlogic switches with Palo VIC card as well as ESXi using SAN for guest shared datastore, no issues.

Do you have any details on how this worked? I setup 2 HBA fc0 and fc1. I set the Boot Order to boot from CD then SAN

From the Switch we see the San Boot target

The server boots, I select the CDrom (no SAN selection appears, but thats OK we have to get it on the SAN LUN first)

ESXi boots from CDrom and no option to install to SAN.

Where am I going wrong?

Hi Harold,

What is the Remote LSILogic Logical Volume of 136+ GBs?

Have you checked the FC path connectivity?

* Seeing that your boot policy is utilizing FC0, can you verify that the WWPN of FC0 has FLOGI'ed into the Fabric Interconnect. Verify with

   UCS-A(nxos)# show npv flogi-table

* If you see that it is logged into the Fabric Interconnect, then check the upstream QLogic 5200 device.

* Not sure about a QLogic 5200 switch, I would assume that they would allow you to check the FLOGI database, but on an MDS it would be as follows:

   MDS# show flogi database

* Have a look around for something similar to this on the QLogic 5200 device and verify that you see the WWPN of FC0 has logged in.

* Also, ensure that the target WWPN of the storage array has logged into the QLogic 5200 switch

* Additionally, verify that the zone is correctly setup to allow the WWPN of FC0 and the target WWPN of the Storage Array to communicate with each other. Ensure the correct WWPNs are specified. Many times people accidently mis-type the WWPN

There are a bunch of things that you can check to verify your setup. Best would be to answer the above questions and also review the UCS Troubleshooting SAN Boot and SAN Connectivity Issues document. It has a whole checklist that you can work through to try identify where the issue is.

Let me know how you go.

Regards,

Michael

I'm getting close to having it up. On the Qlogic switch I created a zone usydoz77, is it correct? 0b0b is the fc1 (I changed to fc1) WWN. The 73c0 is the cisco adapter from UCS

When I boot to install Cisco ESXi  with the Boot from SAN Boot order, I do not see the LUN as discribed on intial post. I have to change to a Local Disk policy then I see the Disk and the LUN. I choose the LUN to install to, let the install finish, shutdown the system, change to Boot from SAN, and it presents a blank screen. Additionly I do not see the flogi for my WWN. Any help would be appreciated.

Ok you lost/confused me there at the end.  You stated that you followed our directions and went as far as installing an OS to that LUN then you changed the boot order and had issues.  WHY did you change the boot order again ??  You should have a working boot policy and local disk policy implemented that allowed you to boot up initially, be seen in flogi database, able to view the LUN as a remote device to install to when you boot an .iso image on virtual media.  Then you install the OS and reboot the server and the ONLY thing you should now change is unchecking the "Map" checkbox if needed for mapping the virtual media.  You made a change that broke the correct configuration, why and what was it.  As explained before you shouldnt' use "diskless" boot with boot from san for several reasons most of which I dont agree with so dont get me going on here, offlice would be better.  Your boot policy shouldn't ever change.  I like to create a "SAN-BOOT" policy boot order policy to implement, using vHBA templates I create prior, with known working SAN Boot Targets onfigured in all 4 slots, vHBA0 Primary and Secondary as PRIMARY SAN Target, and vHBA1 as SECONDARY SAN Target will also have a Pri and Secondary SAN Target WWN configured there to ensure redundancy of pri and sec on both the A and B Fabrics.  So that covered local disk policy is important as SAN LUN is sees as a disk  your policy cannot contain "No Local Storage" which is contained in the "diskless" policy in UCSM by default.  I have seen this actually power off or at least disable adapters on MB.  Those devices arent manageble via the BIOS natively or via UCSM BIOS policy, but you can apply that and watch the FSM and see it happen, and then in the BIOS you will no longer see any hdd's that may have been attached be visible nor will hdd boot be available in BIOS boot  screen, so not good.  If there are disks installed create a SAN with "any configuration" or use the any config for now.  Again you should eject your local hard drives but if using ESXi 4.1 U1 you can leave them.  For your level in SAN boot and UCS I would certainly EJECT them for now, Boot it up with all this discussed, see the LUN, ensure LUN is ZERO and install to LUN with .iso as virutal media. by ensuring your boot order has cd-rom first always, then SAN via local storage adapter, then vHBA's and targets discussed prior.  Dont allow local hdd devices after the SAN.  Install OS, reboot and it should boot from SAN if followed as is.  Put an IP address on the proper interface and vlan # if needed you can access ESXi with vcenter.  Now log into that IP and see what is viewable using the Storage and Storage Adapters screens and tools in ESXi on that blade with those SAN settings you are aware of for that new zone.  Add another LUN to that zone and refresh the storage and make sure you can see it and it's the proper size and read/write acces is there.  Then before moving on, I would do this exact process on a 2nd blade and boot it up the same.  Ensure that the LUN 0 boot for this first blade is NOT viewable by the 2nd blade you're testing with.  After getting this zoned, 2nd biggest issue I see is masking it off properly to LUN 0 for the host, and then ensuring that other hosts also see ONLY their respective LUN 0, NOT the other one or you will overwrite the data and boot sector, or corrupt it at the very least.  If 2nd one doesnt see the LUN 0 bootable to the first, you are good to go, emulate for each going forward.  When 2nd one is running ESXi again using client or vcenter make sure 2nd one can see the same LUN as the first one for shared storage among them both, with this you have covered all bases needed for basic ESXi booting from SAN and shared storage for VM's and vmotion capabilities.

  SAME working SAN boot policy and local disk policies can be applied to SP template to spin many off, or at least included in new Service Profiles for same type san boot machines, same targets and all, will work fine.

  ALSO so now all this is done, and ESXi is 4.1 Update 1 or better, you can power down the blade and insert the hard drives again ejected during san boot install.  A fix in update 1 allows local storage while booting from san, but having them ejected during install even if installing 4.1 U1 ensure smoother boot when not sure on setup and san zoning by removing many variables.  Adding back later like I said here nsures OS brings them properly also with SAN targets having lower numbers, hdd's higher which always matters at the PCIe level where those boot order preferences actually dont matter as you'll see per the notes on the UCSM page when viewing, PCIe order rules.

Good Luck and let me know.

I am using a Boot Policy SAB Boot

When boot the CRrom, Cisco ESXi, I can install to Hitachi LUN

Details

The install finshies. At this point I would expect to be able to look at the LUN via vCenter and see the install files. It is not there.

The flogi shows

Do I have the correct SAN target? 20:00:00:25:B5:01:0B:0B 

Or do I us the WWNN, 20:00:00:25:B5:01:0B:0A, from the storage tab.

The SAN boot target is just that, and your target is the device your LUN is configured on, so the WWN of the array you are configuring and advertising to the zone.  That WWN could be just the port that the array is plugged into on the switch, but that is your hosts "target".   It's the WWPN not the WWNN that you want to use to put into your zone so zone the 20:00:00:25:B5:01:  0A:0E and 0B:0B it looks like on your fc0 and fc1 vhba screen.  So no it's not correct, you are using your fc1 WWPN as the SAN target, SAN target is where your LUN is, your vHBA's are TARGETING those primary and seconday SAN targets so they should ALL belong to the storage device, the zone fabric must allow those I just typed onto the fabric, then to see that LUN and see it as 0.

  Everything you see 20:00:00:25:B5 is Cisco , so those cannot be SAN targets, you need your SAN storage device's WWN for a target and that means he must be in zone as well so they see each other.

You are on the right track, you wont see an option to boot from san but both sites I just installed with these switches I went thru this same scenario for hours until they got it right.  I wont touch upon the boot process because you are seeing the WWN hit the fabric so that should be fine.  Then after adding the WWN to a zone or a storage group he's alllowed to joing the fabric.  At that time you need to ensure there is a LUN in the same group so he can see it, and it needs to be masked so only he sees it and sees it as LUN 0.  ESXi wants LUN 0 and it's really picky about those local drives being visible, you need to eject them to be safe.  But when the LUN is visible it will show on that screen there under remote, but just like a hard drive.  I bet you arent using the Cisco VMware image, I cannot see what version you are using but download the Cisco OEM image from VMware.  After I started using that with the vic card, my boot from san issues went away.   When your LUN is visible in the select a disk window just choose it to install to like a local disk.

So if SAN boot with ESXi and VIC, use cisco oem image of esxi, eject local hard drives, do NOT use diskless boot policy, use "anything" and eject at least until it's running, then there is a fix for esxi so local drives can exist, but if using for Cisco UC it's not supported.  Lastly it's the LUN that needs to be visible....that's it.  my money is on vmware .iso image due to drivers if your SAN array is correct..

View solution in original post

Hi Harold,

From Your attachment I understand that you have entered the LUN ID is 9. UCS detects the LUN for boot as ID 0.

Please change lun id as 0.

Also I saw that taget WWN seems to a cisco's vHBA's WWPn. Idealy this must be a WWPN of storage.

Please apply Storage's WWPN and zone it accordinly.

Ram


I get it. So here is what I have:

UCS Boot Order where 50:06:0E:80:10:4B:5B:89 is one of the Hitachi SAN interfaces

From the QLogic switch the active Zone Set is Orica_denver where the Zone usydoz77 has all the Hitachi interfaces and the WWPN of the Storage vHBA

When the blade is booted it display a blank black window. The cusor moves down several lines but does not display anything.

Where do I go from here?

Whenever I see the blinking cursor in black window it's been the UCS side where the boot order is not executing.  Last 2 times my customers had that issue, they had the boot order setup correctly but the local storage policy was setup wrong.  I ejected their local hard drives and set the storage policy to "any" and it started booting up normally.  Then if SAN issues still exist they can be properly troubleshot.  My guess is if you are seeing the same thing then the hba WWN is not being seen in the database of the switch.  I have seen when they set the local policy to diskless, when I change it to any you will see the local adapters power back up in the FSM window, those being powered down or disabled is why it cannot boot.

It's working!!!

Most everything you posted worked except for one thing. For some reason beteen UCS, Qlogic switches, and Hitachi 2100 SAN, I could not use LUN 0 for Boot. I had to use the LUN number that was defined in the Hitachi SAN.

I have spent hours on the phone with Cisco, Hitachi and VMware and nothing they said helped. I fact Cisco is still saying that is not setup correctly.

Thanks for all your help. I would have never gotten it to work.

Trust me that single piece is not wrong. Glad you have it working but its not because of that. There are many dependancies on lun 0 being just that for boot, too many to list.

Congrats

Sent from Cisco Technical Support iPhone App

Create
Recognize Your Peers
Content for Community-Ad