Boot from SAN installation procedure

Hi Kiketap,

It's fairly straight forward, but there are a number of steps:
On the Clariion:
1. Firstly, you'll need to bind your LUN's..
2. As the host is not running, you'll need to manually register the Hosts HBA WWN's - don't forget to set the correct failover mode for your configuration.
3. Now put the new host and the LUN's into a storage group.
4. Check the storage group properties, and note the "HLU" of your desired boot LUN

On your switches, create and activate the zoning as per usual. Make sure you have a path to SPA and SPB.

Now on the server, the process will depend on the HBA you are using, but the basic process is as follows:

1. Ensure your HBA has the correct firmware and boot bios installed.

2. Boot the server and enter the HBA configuration utility to enable the boot bios.

3. At the same time, set the topology to "fabric" (assuming you are using a switch - you have to set this here as when the system first boots, the HBA driver has not yet loaded to tell the HBA that it's in a switched fabric environment. Sometimes you need to reboot here before the next step works as the HBA may be set to FC-AL. Also, some HBA's refer to "fabric" as "Point to Point"

4. Use the HBA utility again to discover and and then Configure the Boot LUN's - you'll need to use that HLU you noted earlier to identify the correct LUN. Take a look at the instructions in the docs below.
If you have a single HBA configuration, ensure you discover the boot LUN on both SP's (ie: configure 2 boot lun's) otherwise if the LUN tresspasses, the host won't boot.

If you have 2 HBA's, then you can discover the boot LUN on both HBA'a and either 1 SP per HBA or configure both SP's per HBA (ie: 4 boot LUN's) - just follow your regular zoning/setup to make life easy.

Now once that's finally, done, you may need to reboot again, and enter the server's BIOS configuration - usually in there somewher is the hard-disk boot order, and if you've got it all sorted correctly you should see the HBA listed. Promote the entry for your HBA to the top.. You're now finally ready to start the WIndows install...

Do this as usual - make sure you have the appropriate HBA driver ready (you'll need to hit (F6 ?) when it prompts at the beginning of the install)

Once Windows is installed, you'll probably need to install the HBA driver again and the HBA utilities, set the HBA settings for your environment, Install Powerpath of course, Oh and the Navi agent, CLI... I'm sure you've got the rest sorted.

The procedure is documented for emulex HBA's here: http://www.emulex.com/emc/support/pdfs/win.pdf

and for Qlogic HBA's here: http://download.qlogic.com/drivers/42180/QLogic_Windows.pdf

And you could also take a look at the Host Connectivity Guide for Windows which is here https://elabnavigator.emc.com/emcpubs/elab/config_guide/windows.pdf

Regards,
Glen.

I have been through this before... and till now the procedure generator from EMC does not include any "Boot from SAN" procedures, you will have to get this from the HBA vendor.

But of course GlenH has already answered your question:) and I just wanted to clarify this point.

Regards.

Glen,

Awesome write up man. I would copyright it though before EMC does

Regards,
Shmac

GlenH,

i have a question about step 4. I had a box with 2 hbas, one hba was zoned to SPA and another hba was zoned to SPB. Even thought i had boot lun configured on both hba's , there were instances where the server could not find the boot drive unless i went into Navisphere and trespassed it to another SP. It's like from boot to boot it would rescan different hba ..obviously at this point it cant trespass the lun so it would just halt the boot process. Any ideas?

Can Powerpath see more paths to the boot up drive once the server is up and running? This would assist in the case that the assigned boot up port on the san would happen to fail.

Thanks,
CM.

On PowerLink, you can find the Host Connectivity Guides for each operating system. Each guide has information about booting from SAN. Generally, you need to go to the EMC specific section on the HBA vendors WEB pages and get the installation guide for the HBA - there is a section about setting up booting from SAN.

glenK

Hi Glen,

Great Step by step guide...had to learn it the hard way!....I have one doubt, would powerpath help the server recover from HBA failure in a boot from San setup without the need to reboot the server? My setup is ..ive got blade servers connected in a fibre switched network. Each blade has 2 HBAs and both the HBAs are connected to each SP. Now without powerpath, if one HBA / link fails, I have to reboot the server and it uses the second path to the SAN.

Waiting for your reply.

Thanks
Sai

yes, PowerPath should allow the host to keep on running. I yanked a cable from one of my system and the only thing that i saw was a 5 seconds delay...like the box froze or something, but then it kept on running without requiring a reboot.

thanks dynamox.

I will make sure i get powerpath for all my servers, when i move into the production deployment.

Cheers
Sai

maybe the auto-tresspass flag needs to be turned on for the luns.
auto-tresspass would allow the sp to trespass the lun on its own when a request is sent to a port (or sp) that does not own the LUN.
is the setup still in place so that you can check that out?

nah ..i want Powerpath to take care of failover.

Remember that PowerPath starts up after the boot.

glenk

i realize that ...but i do not want the array to initiate failover once the box is up ...that's what PowerPath is for.

autotrespass will not affect powerpath behaviour...

powerpath on clariion behaves in a different way. powerpath keeps track of the paths that are going to owning SP. these paths are not used to request i/o unless primary paths fail. you can check this by powerpath gui when there is large i/o going on. secondary paths will show 0 in stats.

when using autotrespass, sp will not keep checking paths like powerpath does. hence array will not initiate trespass when a path fails (cable breaks, port fails, hba fails etc...) trespass is initiated only when there is an explicit i/o coming for that lun on peer SP.

so in your case, at boot time, trespass will help host pulling the lun but after powerpath takes over, it wont interfere with powerpath...

when you mention auto-trespass ..do you mean auto assign ?

is this unique situation because it's booting from SAN? Here is excerpt from Navisphere Manager Admin guide:

If you are running failover software on a server connected to the
LUNs in a storage system, you must disable auto assignment for all
LUNs that you want the software to fail over when an SP fails. In this
situation, the failover software, not auto assign, controls ownership
of the LUN in a storage system with two SPs.