I have a two-node P4 Xeon 32 bit RedHat 4 AS cluster using Oracle CRS as its clusterware. I utilize dual qla2200 HBAs to talk through a Brocade 2250 switch to two-JBOD arrays, a Nexstor 18f and a Nexstor 8f, both loaded with 72 gig 10K Seagates. I do not use a RAID controller on either JBOD.

I have a OCFS2 setup on two of the JBOD drives that contain the voting and config files for the cluster software, the cluster comes up fine and is configured using multipath and DM to use active-active through the two controllers on each host.

I am configuring Oracle ASM. Using the oracleasm tool I can query and see the ASM disks on both of the JBODs. All of the ASM disks show up as multipathed using dminfo and multipath -V3 -ll in an active-active configuration.

When I start the ASM environment on Node A, Node B ASM will not start, if I start the ASM first on Node B then Node A cannot start ASM. There appears to be a permission/global sharing/scsi reservation issue, but I'll be damned if I can find it.

Anyone come up against this before? This is with Oracle11g, release 1, my kernel is 2.6.9-67.0.15 ELsmp #1.

Thanks

The issue was't with multipathing, it was with having two RAC interconnects defined, seems ASM can't handle that from the install so it was failing, I took one of the interconnects out of the configuraiton file and poof, everything works (interconnects are ethernet or other network paths between the two or more computers in the cluster and have nothing to do with multipatning.)