HBA instance No.

[Girish_17]

Hi Friends,

I would like to see the shared storage from two systems with same device file names.
From one node i am able to see it as c9txdx and from other node it is c8txdx.

The difference in the instance no is creating problem for me. I want both of them to be same e.g. it should be c9txdx on both systems.

Since this are the remote servers i really can't play with the HBA (putting in different slots or so) and also i don't want to create soft links for some reasons.

Is possible to do the required change in either of the nodes to make thier instance numbers same. I won't mind rebooting the servers.

Thanks a lot for your suggestions and comments.

Regards,
Girish

[A. Clay Stephenson · Accepted solution]

The fix is rather simple although it should not be used to change boot devices:

ioscan -f | grep -e "INTERFACE" -e "DEVICE" | grep -v -i "target" | awk '{printf("%-20.20s %-12.12s %3d\n",$3,$1,($2 + 0))}' > /myfile

Using vi, edit /myfile and change the lan instances or disk devices as you wish but make sure than no instance numbers are duplicated.

mv /stand/ioconfig /stand/ioconfig.safe
mv /etc/ioconfig /etc/ioconfig.safe
shutdown -r -y 0

The system will stop the reboot at the ioinitrc prompt because of the missing files:
Issue this command:
/sbin/ioinit -c
/sbin/ioinit -f /myfile -r

You may get warnings like "input identical to kernel" -- these are normal and expected and may be safely ignored. The "-r" triggers a reboot.

You configurations should be correct now. Man ioinit for details.

---------------------------------------

The other way to approach your problem is to rethink it. Rather than connecting to /dev/rdsk/c8t5d0, for example, your application connects to /xxx/yyy/rdisk1 and you create symbolic links to the actual device nodes.

[Girish_17]

Thanks a lot clay .........it worked.

But i need some clarification.

In My file i had something like:
0/3/1/0.1.11.255.14 ext_bus 5
0/3/1/0.1.11.255.14. disk 6
0/3/1/0.1.11.255.14. disk 7
0/3/1/0.1.11.255.14. disk 8
0/3/1/0.1.11.255.14. disk 9
0/3/1/0.1.11.255.14. disk 10
0/3/1/0.1.11.255.14. disk 11
0/3/1/0.1.11.255.14. disk 12
0/3/1/0.1.11.255.14. disk 13

I changed it to:
0/3/1/0.1.11.255.14 ext_bus 4
0/3/1/0.1.11.255.14. disk 5
0/3/1/0.1.11.255.14. disk 6
0/3/1/0.1.11.255.14. disk 7
0/3/1/0.1.11.255.14. disk 8
0/3/1/0.1.11.255.14. disk 9
0/3/1/0.1.11.255.14. disk 10
0/3/1/0.1.11.255.14. disk 11
0/3/1/0.1.11.255.14. disk 12

i wanted to change from /dev/dsk/c5txdx to /dev/dsk/c4txdx.

So my question is whether i have to change entire order with +1 as in my case? (it actually worked that way only) or i have change just the first entry to 4 (This is what i was expecting).

Secondly when i reboot i did /sbin/ioinit -c and then /sbin/ioinit -f /myfile -r but here it didn't reboot, it gave me back the prompt and then i used /sbin/reboot.

After reboot i was able to see the change but got confused with the way it worked.

Waiting for your valuable guidance.

Regards,
Girish

[Girish_17]

Waiting for the response...........
Thanks
Girish

[A. Clay Stephenson]

No, you do not have to "ripple" all the changes BUT if there were already a c4 and you wanted to change your existing c5 to c4 then you would have to change the existing c4 to some other (available) instance and then change your c5 to instance 4. If you do a man ioinit you will see that -r doesn't always trigger a reboot though in practice, it usually does --- especially if you don't make any mistakes.

[Girish_17]

thanks clay........