Some site uses double hosts(SUSE Linux9 sp3)+ double FC switch+S3100 storage to make cluster. When the two optical fibers between one of the hosts and the storage are in fault, VCS cluster does not switch automatically, and lead to operation cut.
There is “ [RAIDarray.mpp]DBArray:0/1:any:0: mppVhba command 0 failure - RESERVATION CONFLICT. vcmnd SN 9104566” error in message log of the host.
Cancel the disk reservation configuration in VCS.
The problem is lead by the collision between VCS disk reservation strategy and RDAC multipathing software. If VCS is configured disk reservation strategy, it will collide with RDAC, and lead to the problem.
Don’t configure disk reservation when there is multipathing software.