How does loop # change to effect MCSG environment?

[Jeongbae Min]

Hi,

We need more connection to EMC storage and we want to replace 2 ports FA to 4 ports FA. It will change loop #. and I heard that it will change controller #. then I think it will change all device information to attached that EMC port.

What would be a procedure to replace EMC FA board?

Thanks

[Steven E. Protter]

It is possible if not probable that this procedure will change the device identification in /dev/dsk

I would make_tape_recovery or make_net_recovery, and consider vgexport before the switch.

If the numbers change, you'll be able to vgimport after the fact and have an easier time with the transition.

Don't have a procedure for you. Sorry.

SEP

[Sridhar Bhaskarla]

Hi,

Change of device files will affect your lock disks. So, it is cleaner to re-apply the serviceguard configuration. There are unsupported tools available to change the lock disk.

1. Halt the packages and cluster.
2. Generate different kinds of map files. Copy them onto the failover box.

vgexport -v -p -s -m /tmp/vg01.s.map vg01
vgexport -v -p -m /tmp/vg01.map -f /tmp/vg01.disks vg01
ll /dev/*/group > /tmp/vg.groups

3. Deactivate the volume groups on both the nodes.

vgchange -c n vg01
vgchange -a n vg01
vgexport vg01

4. Make sure you don't have any reference to the se device files in /etc/lvmtab

strings /etc/lvmtab |grep -e 'cx' -e "cy"

5. Complete your EMC work.

6. Run ioscan followed by insf

ioscan -f
insf

7. Import the VGs back on both the systems.

vgimport -v -s -m /tmp/vg01.s.map vg01

This should automatically import the VGs. Or you can use -f option but you have to modify vg01.disks file to change the controller numbers.

8. Activate the VGs on one system.

vgchange -a y vg01

9. Edit /etc/cmcluster/your_cluster.ascii file and change the lock disks. Distribute this file across both nodes.

10. Apply the configuration (cluster and package)

cmapplyconf -C /etc/cmcluster/your_cluster.ascii

11. Start the cluster
cmruncl

12. Apply the package configuratoin
cmapplyconf -P /etc/cmcluster/your_app/your_app.conf

13. Start the package.

14. Verify /var/adm/syslog/syslog.log for any errors.


-Sri

[Tim D Fulford]

As always Sri's reply is excellent.

Point 9 though is NOT necessary IF (and only IF) you do not have any lock disks. If you are running a 4-node cluster or more you can skip 9.

Regards

Tim