Testing the rdac failover mode, I notice that it should failed IO instead of hangs it when we unmapped volumes on a SAN. Same bug as here : http://patchwork.kernel.org/patch/54943/ Reproducible: Always Steps to Reproduce: 1. Mount a volume 2. Do IOs operation on it (cat /dev/urandom > test) 3. unmapped the volume on SAN Actual Results: The process hangs instead of failed. Shell is not recoverable... If we do Raid1 between 2 volumes, if one failed the IOs are hang... so it's impossible to use the second volumes..... Expected Results: The volumes should be marked as failed and IOs too. So Raid1 will do the rest to recover from the situation.
Fixed in tree: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=e78d65c9e9ffa54a6bece600c8d09881361c4889