Adding new LUNs

[The_Cina]

Hello guys,

A quick question on agile device file name assignement.. :

If I decide to add additional LUNs to an HP-UX11.31 host, what would be the number reported on the device file in agile view format /dev/disk/disk### ?

To better explain the question, let's suppose I have alredy a LUN (LUN# = 10) assigned to the host, and it has its device file /dev/disk/disk390.

If now I add a new LUN to the host, the device file name associated will be the next to the previous one, i.e. /dev/disk/disk391 ? Or it will somehow be dependent from the LUN# specified on the Storage?

Thank you in advance for your help,

[Deepak Kr]

Device file name depends on multiple things.

HBA Card, port, san switch port, number of san switched and finally storage box.

Anyways you always run ioscan following by insf to see what is the default name came once you allocated lun to a hp box.

[Duncan Edmonstone · Accepted solution]

Actually for agile DSFs in 11iv3, usually you'll find it will be the next device file up (disk391 in your example). But it won't always be.

Device file names are related to the instance of that device type stored in the ioconfig file. You can see instance numbers for devices currently attached to the system by using ioscan. To see devices of type esdisk (which is the device driver for the agile DSFs), you would run:

ioscan -kfnNd esdisk

The 2nd column (labelled I) shows the instance of the device, and unless you have been doing some remapping you should find that the instance number is always consistent with the number in the DSF name.

Of course this gets more complex if LUNs have been presented and then later removed, as the entries aren't removed from the ioconfig file. So in your example you might find that disk391 is the DSF for the next presented LUN, but if you then un-presented that LUN and then presented and entirely new LUN a few weeks later, that would be disk392 even though from the output of ioscan you might think it would be disk391. This is because those "stale" instances aren't removed from the ioconfig file (you wouldn't want to or you'd have a nightmare every time a hardware component failed).

Unfortunately the ioconfig file is in binary format, so you can't look at it under normal circumstances. If you have a support contract then you *might* be able to persuade the response centre to give you a tool that will allow you to look at the contents of the ioconfig file (if you're that interested).


HTH

Duncan

[The_Cina]

Thanks Duncan, Straight to the point!

The case you have reported (deleting the old LUN and adding a new one) is actually what I'll be facing soon, and I'm already scratching my head on it...

In addition I have the following variables:
- This is a Cluster environment
- On the second node (don't ask me why) Instance # and device file name are shitfed down by 1 I #=391 --> device file name /dev/disk/disk390.

Perhaps, as you suggested I should open a ticket with HP and before add/remove LUNs fix this Instance number issue.

Thanks