how can iscsi initiator map a LUN to a specific drive

Discussion in 'Server Operation' started by iinfi, Jun 7, 2009.

  1. iinfi

    iinfi New Member

    i am a novice Linux admin and our office does not have a storage box. i am using openfiler as a storage to practice different stuff which need common storage.

    i am practicing clustering with RHEL5.3 and facing an issue with mapping common storage with the RHEL 5.3 nodes.
    i tried to create a 2 node cluster and here is what i am facing.
    i created 3 LUNs on my openfiler and ran the following commands on both nodes.
    iscsiadm --mode discovery --type sendtargets --portal 192.168.1.10

    when i ran fdisk -l on each node i find that the LUNs are mapped but they are not in correct sequence
    on node 1 -- LUN1 -- /dev/sdb
    on node 1 -- LUN2 -- /dev/sdc
    on node 1 -- LUN3 -- /dev/sdd

    but on node 2
    on node 2 -- LUN1 -- /dev/sdb
    on node 2 -- LUN3 -- /dev/sdc
    on node 2 -- LUN2 -- /dev/sdd

    /dev/sdc and /dev/sdd are swapped. i am not too sure whether things would work right when i configure any application on the cluster.

    is there a way to map a LUN to a specific drive? like map LUN 1 to /dev/sdb and LUN2 to /dev/sdc and so on so that consistency can be maintained on all nodes.


    there is a procedure given here but i am not able to follow it correctly and it doesnt seem to work. is there any other way around for the same? mapping drives through UUID is an option which someone suggested here but if i need raw devices this might not work.
    i am novice linux fellow so please excuse if the above query looks too stupid.
     
  2. falko

    falko Super Moderator ISPConfig Developer

    I'm not sure how you can map a LUN to a specific drive...
     
  3. iinfi

    iinfi New Member

    well,
    when i run the command on one node
    iscsiadm --mode discovery --type sendtargets --portal 192.168.1.10

    and restart the iscsi service

    i get 3 additional drives.
    when i do the same on node2 i get 3 additional drives.
    suppose on node1

    sdb --- 10MB
    sdc --- 15MB
    sdd --- 20MB

    on node2 it becomes
    sdb --- 15MB
    sdc --- 10MB
    sdd --- 20MB

    and the same changes on reboot of the node. well am i being clear?

    thanks for your reply.
     

Share This Page