This issue appears on the configuration of a two-node Oracle Solaris 3.3 3/13 cluster plus the Oracle Solaris 10 1/13 recommended patch set and quorum devices using the SCSI-2 protocol, configured as Oracle VM Server for SPARC guest domains serviced by Oracle Solaris 11.3 I/O domains.
If one node leaves the cluster, either due to a node halt, panic, reboot event, or loss of cluster interconnect (that is, split brain), delayed SCSI-2 access to a quorum device might lead to a panic on the other cluster node, producing an error similar to this one:
panic[cpu13]/thread=30011df80e0: CMM: Unable to acquire the quorum device.
Workaround: On guest domains running Oracle Solaris 10 1/13 plus recommended patches, configure the quorum device to use the SCSI-3 protocol.
Recovery: If the workaround is unsuccessful, reboot the server. Collect any panic core dumps and contact your service representative.