Cloning HPUX host attached to EMC Clariion disk array

[Mark Coleman_1]

Hi everyone,
Just wanted to post a message so hopefully someone else won't get bit like I did last night. If you ever clone a system that is connected to an EMC Clariion, or any HPUX host that runs EMC's Navisphere agent, before you attach the cloned host to the array (or even boot it up on the network), Please go to the /etc/log directory and delete the HostIdFile.txt. Navisphere Agent creates this file to uniquely identify the host to the disk array. When a host is cloned this file goes along for the ride and does not get reset at every boot. If you allow Navisphere Agent to start up, it will identify 2 hosts to the array with the same information, which causes the array to only display 1 host in Navisphere manager. At 2 am this can cause system administrator panic because you won't be able to add the cloned machine to any storage groups. Deleting the HostIdFile.txt will cause the agent to pick up the new ip address of the cloned machine and come up properly. You also have to call EMC support to get them to restart the management server on each service processor of the array.
Hope I saved someone some pain and aggravation in the future. Who puts config files in /etc/log anyway? I love my EMC, but this was a pain.

Mark

[isaac_loven]

Mark,
Thanks for the info. We are installing A Clariion in the next month, so your info is appreciated.
Isaac

[Brian Butscher]

Thanks for the info Mark! Unfortunately, I didn't see your post until this morning as I just cloned one K460 to another that is attached to a CX600. What a pain! We have quick loop adapters, A3404A, and they don't play well in the SAN box (budget contraints, whay can I say! You are so correct about the /etc/log/HostIdFile.txt. With the loop adapters, it dropped the connection to the first K box, down goes the database! EMC support did recommend that on the first K box to change the 'pvtimeout' value from the default (30 seconds) to 90, I changed it to 120, and to make sure the 'lvtimeout' value is set to the default of '0' (wait forever). The 'pvtimeout' can be changed with 'pvchange -t 90' /dev/dsk/cXtXdX' for each physical volume on the Clariion and this can be done at anytime while the system is running. To change the 'lvtimeout' first check the value with 'lvdisplay /dev/vgXX/lvolXXX'. This value can be changed with 'lvchange -t 0 /dev/vgXX/lvolXXXX'. I'm still having connectivity problems with both K boxes when they are connected to the Clariion. I hope to have some resolution for this today. The Clariion leaves a lot to be desired when using it with quick loop.

Regards,

Brian