Why efi_ls does not work with fiber channel disk?

[Kim Tran_4]

efi_ls works as expected with three internal disks. When I try it with a fiber boot disk, it complains "invalid efi_device". Attached is my ioscan:

iatest4:/>ioscan -fnC disk
Class I H/W Path Driver S/W State H/W Type Description
=============================================================================
disk 0 0/0/2/0.0.0.0 sdisk CLAIMED DEVICE TEAC DW-28E
/dev/dsk/c0t0d0 /dev/rdsk/c0t0d0
disk 1 0/1/1/0.0.0 sdisk CLAIMED DEVICE HP 36.4GST336706LC
/dev/dsk/c2t0d0 /dev/dsk/c2t0d0s2 /dev/rdsk/c2t0d0s1
/dev/dsk/c2t0d0s1 /dev/rdsk/c2t0d0 /dev/rdsk/c2t0d0s2
disk 2 0/1/1/0.1.0 sdisk CLAIMED DEVICE HP 36.4GST336706LC
/dev/dsk/c2t1d0 /dev/dsk/c2t1d0s2 /dev/rdsk/c2t1d0 /dev/rdsk/c2t1d0s2
/dev/dsk/c2t1d0s1 /dev/dsk/c2t1d0s3 /dev/rdsk/c2t1d0s1 /dev/rdsk/c2t1d0s3
disk 3 0/1/1/1.2.0 sdisk CLAIMED DEVICE HP 73.4GATLAS10K3_73_SCA
/dev/dsk/c3t2d0 /dev/dsk/c3t2d0s2 /dev/rdsk/c3t2d0 /dev/rdsk/c3t2d0s2
/dev/dsk/c3t2d0s1 /dev/dsk/c3t2d0s3 /dev/rdsk/c3t2d0s1 /dev/rdsk/c3t2d0s3
disk 4 0/4/1/0.2.10.0.0.0.1 sdisk CLAIMED DEVICE COMPAQ HSV111 (C)COMPAQ
/dev/dsk/c4t0d1 /dev/dsk/c4t0d1s2 /dev/rdsk/c4t0d1 /dev/rdsk/c4t0d1s2
/dev/dsk/c4t0d1s1 /dev/dsk/c4t0d1s3 /dev/rdsk/c4t0d1s1 /dev/rdsk/c4t0d1s3
disk 5 0/4/1/0.2.10.255.0.0.1 sdisk CLAIMED DEVICE COMPAQ HSV111 (C)COMPAQ
/dev/dsk/c5t0d1 /dev/dsk/c5t0d1s2 /dev/rdsk/c5t0d1 /dev/rdsk/c5t0d1s2
/dev/dsk/c5t0d1s1 /dev/dsk/c5t0d1s3 /dev/rdsk/c5t0d1s1 /dev/rdsk/c5t0d1s3

AND THE FAILED ONE IS:
iatest4:/>efi_ls -d /dev/rdsk/c4d0d1s1
efi_ls: open device failed(errno = 2)

I really appreciate your help.

[Kevin Wright]

you don't have an efi partition on the disks.

[whiteknight · Accepted solution]

You may create EFI,OS and HPSP partition using idisk(1M).

(1) Create description file. (/tmp/partitionfile)

3
EFI 500MB
HPUX 100% or XXXXMB
HPSP 400MB

- HPSP partition must locate after OS partition.
- You may specify OS partition size either % or actual size.
- EFI partition size and HPSP partition size may up to your
disk size.
So you should check original disk partition size.

(2) Create partitions
# idisk -wf /tmp/partitionfile /dev/rdsk/c4t0d1

(3) Confirm
# idisk -p /dev/rdsk/c4t0d1

see detail:-
http://docs.hp.com/en/5187-2725/ch06s06.html


WK
please assign points

[Torsten.]

typo

iatest4:/>efi_ls -d /dev/rdsk/c4d0d1s1
(2x "d")

correct

iatest4:/>efi_ls -d /dev/rdsk/c4t0d1s1

[Kim Tran_4]

Thanks for pointing out the typo. The correct one does not work either:

iatest4:/>efi_ls -d /dev/dsk/c4t0d1s1
efi_ls: invalid efi device

How do we know that this disk does not have efi partition? The way ioscan shows similar info for all disks, and look like they all have ...s1, ...s2, ...s3? By the way, the disk /dev/dsk/c4t0d1s1 is a LINUX boot disk. Thanks

[Kim Tran_4]

Thanks for pointing out the typo. The correct one does not work either:

iatest4:/>efi_ls -d /dev/rdsk/c4t0d1s1
efi_ls: invalid efi device

How do we know that this disk does not have efi partition? The way ioscan shows similar info for all disks, and look like they all have ...s1, ...s2, ...s3? By the way, the disk /dev/dsk/c4t0d1s1 is a LINUX boot disk. Thanks

[Torsten.]

invalid - looks like there is no FAT filesystem on this partition.

For a hp-ux related EFI partition you would create it with

mkboot -e -l /dev/rdsk/cxtydz

but this will give the OS loader for HP-UX, not linux.

You should look for a tool from your linux distribution to achieve this.

[Steven E. Protter]

Shalom,

I always recommend against booting off fiber. If the SAN or any component goes, you lose the whole system and have nothing to run diagnostics with.

Only certain SAN disks may be supported by EFI. You might want to try and update the firmware on your box.

SEP

[Torsten.]

IMHO, booting from SAN is not the problem, because it is very redundant - maybe more than an internal disk.

If your box is a test system sometimes booted with linux sometimes with hp-ux, you should simply ignore the linux volumes.

All other actions may lead to strange results.

Look, the vdisk on the EVA is the same, but the host mode is not hp-ux. For this reason you see it twice here:

0/4/1/0.2.10.0.0.0.1 c4t0d1

0/4/1/0.2.10.255.0.0.1 c5t0d1

The reason is the addressing mode depending on the host mode - wrong setting for hp-ux.

Normally you would not present the vdisk to hp-ux, but here simply ignore it.

[TwoProc]

The number-one-fan-of-booting-from-a-san
is Nelson Caparosso. I used to have clustered systems booting from a san and have gotten off of it, and won't go back (no rehab needed).

Nelson, can you weigh in on this?

[Kim Tran_4]

Booting from SAN is very efficient in a test enviroment. What we are trying to do is to have a master system, which continuously ignites OS of different versions (11.23, 11.31 for HPUX, SuSe9, 10, Redhat 5.xx, 4.xx for LINUX) from SAN. Whenever someone request a clean system for testing, we just attached the SAN boot disk to the system, message it a little bit, and gives the requestor the new system within half an hour. So far we are ok with either all HPUX, or all LINUX. Now we would like to automate the switching between HPUX and lINUX, and a whole new world is open to us :)

[Kim Tran_4]

Thanks for all your help. The problems is the boot partition of SuSe is s2, not s1 like others. Now it works:


iatest4:/>efi_ls -d /dev/rdsk/c4t0d1s2 /efi
FileName Last Modified Size
. 10/ 4/2004 0
.. 10/ 4/2004 0
SuSE/ 10/ 4/2004 0
Intel Firmware/ 10/ 4/2004 0

total space 210537984 bytes, free space 203924992 bytes

Again, thanks for all your input.