OCR Problem on losing storage
we've the following setup: AIX5.3TL8SP4, Oracle Clusterware 10.2.0.4, Oracle Database 10.2.0.4, RAC consisting of 1 Partition (p6) in Location "EDV" + 1 Partition (p6) in Location "LWST", Storage 1 DS5k in "EDV" + 1 Storage DS5k in "LWST", Mirroring is done using ASM. We've also mirrored OCR across locations and voting disks (1 in "EDV", 2 in "LWST").
if we loose connection to Storage in "LWST", both nodes crash & reboot (due the loss of quorum reg. voting disks), which is quite expected. if the connection to storage in "LWST" isn't reenabled after the reboot we face the problem, that CRS is not starting up, due to the failing OCR mirror. unluckily we were unable to change OCR config in this state (ocrconfig -replace ocrmirror <failingpath>) awa changing voting disk config (crsctl add css votedisk <newpath> -force & crsctl delete css votedisk <one-of-lwst> force), hence cluster won't start up until storage problem is solved, which is not acceptable.