Hardware path for disk are diffrent across nodes

[Karthik S S]

Hi,

I have two similar A500 with the following configuration.

HP-UX 11.11
1 FC HBA ( Connected to slot 3 )
both machines are connected to VA7110 via FC switch.

The LUNs that are configured on VA are visible on both of these machines. But the problem here is system1 has the device files as,

/dev/dsk/c6txdx

where as system2,
/dev/dsk/c5txdx

Since I am configuring this disks under a cluster with EMS resource monitoring I need the paths to be exactly the same on both the nodes. How do I achieve this??

Pl. Help.

Thanks,
Karthik S S

[Mark Grant · Accepted solution]

You need to look into "ioconfig".

"man ioconfig" will give you al the details you require.

[Karthik S S]

I tried insf -e -C disk but the path doesn't change even after a reboot. These machines are connected to only one FC switch and VA. No other storage/switches involved in this setup.

I am not getting any clue from ioconfig :-(

Thanks
Karthik S S

[melvyn burnard]

Why do you need the device files exacty the same?
Are you refering to serviceguard cluster?
this is not necessary.

[Karthik S S]

I am referring to serviceguard cluster. But the problem here is I am monitoring disk RESOURCES for a package.

package ascii file is configured as,

RESOURCE_NAME /vg/vgsp/pv_pvlink/status/c6t12d7
RESOURCE_POLLING_INTERVAL 60
RESOURCE_START AUTOMATIC
RESOURCE_UP_VALUE = UP

so if the link goes down on the node1 it is supposed failover to node2. But since node2 has the device files starting with c5txdx it wont be able to start the package but gives the following error,

The EMS registrar was unable to contact or locate a monitor for resource /vg/vgsp/pv_pvlink/status/c6t12d7


Pl. help

Thanks,
Karthik S S

[melvyn burnard]

Well if you really want to do this, take a look at the ioinit command. make sure you have an Ignite Recovery backup first, just to be safe

[Karthik S S]

Thank you melvyn,

Can I follow these steps on node2??
(Thanks to sanjay
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=26117)



If you have two identical system with identical hardware on both the systems, configure the drivers as required on both the system and then to make the ioscan output identical from both the systems, try this,

# mv /stand/ioconfig /stand/ioconfig.sav
# mv /etc/ioconfig /etc/ioconfig.sav
# shutdown -ry 0

Due to the missing ioconfig files the system will come to an ioinitrc prompt. Now recreate new ioconfig files from scratch. This prevents you from running into possible assignment conflicts.

(in ioinitrc)# /sbin/ioinit -c


-Karthik S S

[Kenneth Platz]

This should work. Run the following command on system2:

ioinit -f /tmp/ioinit.txt

Where /tmp/ioinit.txt has the following contents:

X.Y.Z ext_bus 6

Now this will assume that you don't already have a controller instance 6 on system2. X.Y.Z is the hardware path of the FCP Array Interface device that controls those disks. Ie, if you did an ioscan -kfn and found the following:

ext_bus 5 0/4/0/0.97.4.19.0 fcparray CLAIMED INTERFACE FCP A
rray Interface
target 6 0/4/0/0.97.4.19.0.0 tgt CLAIMED DEVICE
disk 151 0/4/0/0.97.4.19.0.0.0 sdisk CLAIMED DEVICE EMC
SYMMETRIX
/dev/dsk/c5t0d0 /dev/rdsk/c5t0d0

You would use 0/4/0/0.97.4.19.0 as your hardware path to change disk c5t0d0 to c6t0d0

Also, you will definitely want to have a current make_recovery tape available in case things FDGB (Fall Down Go Boom). Read up on the "ioinit" man page for more details.

[A. Clay Stephenson]

I haven't tried this but it should work. Don't worry about the device assignments just as you don't care what devices a given volume group uses on a given node. I would introduce a level of indirection on each node using symbolic links. Your package might then look for /dev/xxxstatus (or whatever) and each node would have a unique symbolic link linking /dev/xxxstatus to the actual device. This method would also make any future changes easy as well.

[melvyn burnard]

well I would suggest that you use the ioinit command and have it read in a file to create what you wish to have.
From the man page:

To reassign an instance number to a device and class (specified in infile) and reboot the system:

/sbin/ioinit -f infile -r

where infile contains the following:

56.52 scsi 2

56.52 is the h/w_path, scsi is the class_name, and 2 is the instance_#.

[Karthik S S]

Hi All,

Thanks for all those valuable inputs. I am sorry I could not reply you back immediately because for whole of yesterday I could not access the ITRC site :-( ...

I didn't try ioinit but I simply removed the FC card on system 2 and put it on a different slot. And upon reboot new device files were created with c6txdx :-)

Thanks again,
Karthik S S