What’s the best practice for setting up igroups and lun mapping for a cluster?

 

Last night we moved a cluster that uses luns on Netapp (IBM Nseries) to new servers.  The old servers used internal boot disk, while the new use san boot.

 

I had it setup like this (names are different!):

  Igroup: serverA_boot  wwnA1 wwnA2 boot_lun_mapped_in_lunid0

  Igroup: serverB_boot  wwnB1 wwnB2 boot_lun_mapped_in_lunid0

  Igroup: cluster_luns              all_the_cluster_luns_mapped_in_lunid0-149

 

During the cutover, I went to add all 4 wwn’s (wwA1,wwnA2,wwnB1,wwnB2)

into the cluster igroup.  I was swinging the luns from the old

server with internal boot to the new san boot servers and got an

error.  It complained that I couldn’t put the new wwn’s into the

cluster igroup for luns because the boot igroups already had a

lun id 0, and the cluster igroup also had a lun id 0.  That is,

the same wwn can’t map to duplicate lun id addresses.

 

To fix, I simply remapped the cluster lun at id 0 to the next

highest value, then I could add the wwn’s and all was well.

 

Question:  Is this the best way to create a san boot cluster? 

           Is there a better way? 

 

The next time I create a cluster lun It will be interesting what lun id gets assigned – lun id 0 (and get an error), or the next hightest.

 

I also thought of just mapping the cluster luns to both boot igroups, but like that a cluster lun igroup is kind of self documenting.

 

 

 

Thanks for any ideas!

 

Rick