Finding SAN disk device names

[Sean Lee_2]

I'd like to use these SAN disks on Linux, but I don't know device names (/dev/sda, etc.) are empty so I don't know how to fdisk them.

Output from df:
Filesystem 1k-blocks Used Available Use% Mounted on
/dev/cciss/c0d0p3 32938704 4109936 27155560 14% /
/dev/cciss/c0d0p1 51342 26033 22658 54% /boot
none 1544948 0 1544948 0% /dev/shm

Output from dmesg:
Emulex LightPulse FC SCSI/IP 4.20p
!lpfc0:045:Vital Product Data Data: 82 8 0 46
!lpfc0:031:Link Up Event received Data: 1 1 f7 0
scsi0 : Emulex LPFC (LP950) SCSI on PCI bus 06 device 10 irq 15
scsi: unknown type 12
Vendor: COMPAQ Model: MSA1000 Rev: 2.38
Type: Unknown ANSI SCSI revision: 04
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 2.38
Type: Direct-Access ANSI SCSI revision: 04
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 2.38
Type: Direct-Access ANSI SCSI revision: 04
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 2.38
Type: Direct-Access ANSI SCSI revision: 04
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 2.38
Type: Direct-Access ANSI SCSI revision: 04
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 2.38
Type: Direct-Access ANSI SCSI revision: 04
resize_dma_pool: unknown device type 12
resize_dma_pool: unknown device type -1
resize_dma_pool: unknown device type 12

[Colin Topliss]

You don't say which version of Linux you are using. Its sort of important, because there are tools available for SuSE which are not available under RedHat. My systems are running SuSE Linux (8.2, 9.0 and SLES 8).

Still, you can go into /proc/scsi/lpfc - in there you will see a file - its name will be a number. Make a note of that numer. FYI, that file contains the link status - it would be worth checking it to see all is OK. I remember seeing the unknown type error in the early days before I got it working OK - turned out to be an array config problem. Are there any other messages? I had issues with large LUN numbers (Emulex can't handle LUN numbers above 255, and there were problems scanning past LUN 0 in some cases).

Your best bet is to look at /var/log/messages - it will tell you the device numbers right after it loads the driver (at least it did with the LP9000s I tested).

You should see something along the lines of:
Attached SCSI disk sda at scsi channel 0 id 1 lun 0
Attached SCSI disk sdb at scsi channel 0 id 1 lun 2
etc, etc

If you have SuSE linux, try hw-info or sginfo -l

Col.

[Sean Lee_2]

Thanks for your reply.
I use Red Hat Advanced Server 2.1 with e.35 kernel.

I'll try your suggestions.

Thanks again!
Sean

[Stuart Browne]

/proc/scsi/* is your friend. What's there, and what do they say?

[Sean Lee_2]

Yes, it worked to some extent; on one server I can see (below) SAN disks, on the other I cannot see disks, but only SAN devices (as pasted above in the question).
I guess the SAN admin didn't assign these disks to both hosts (or the same "storage group", if that's the right term on HP SAN).


SCSI device sda: 122881440 512-byte hdwr sectors (62915 MB)
sda: sda1
SCSI device sdb: 122881440 512-byte hdwr sectors (62915 MB)
sdb: sdb1
SCSI device sdc: 40955040 512-byte hdwr sectors (20969 MB)
sdc: unknown partition table
SCSI device sdd: 142261440 512-byte hdwr sectors (72838 MB)
sdd: unknown partition table
SCSI device sde: 20481600 512-byte hdwr sectors (10487 MB)
sde: unknown partition table

[Uwe Zessin]

Sean,
I see that you are using an MSA1000 storage array with Emulex adapters - please be aware that this combination is _not_ supported by HP.

You can find the specifications link on the right side of this page:
http://h18000.www1.hp.com/products/storageworks/msa1000/index.html

There is also a newer firmware available via the 'Software, firmware & drivers' link on the same page.


If your MSA1000 has two controllers for redundancy you need two fibre channel adapters in your server(s) and the 'Secure Path' software.
http://h18000.www1.hp.com/products/sanworks/secure-path/index.html

--

In case you are interested in the terminology:
a fibre channel adapter causes a 'connection' object to be created on the MSA1000. It must be assigned the 'Linux' profile to work properly, because there are little differences in communication behaviour between the MSA and different operation systems. Such a connection is then assigned to one or more logical disks so that the server is granted access.

You will find out that the MSA1000 presents a LUN at address 0 that is not a disk device. This is normal. It is a communications channel so that a server can access the array and talk to it even when no disk LUNs have been created.

[Sean Lee_2]

Yes, I realize the fact about the combination and also Secure Path cannot be used in our particular situation.
This is just a demo environment so as long we can see the disks, that's good enough.

The SAN guys managed to make disks available to both servers, so we're done now :-)
(They didn't tell us what exactly they did, so much for knowledge sharing)

Thanks, Uwe.