> 3) sanlock configured (this is evil!) Just out of curiosity, can you please tell more why it is evil? I just found it out after your first post and want to know if there's any gotchas :) > Though it doesn't corrupt data, the I/O performance is < 1% of my > hardwares capability. Hopefully work on buffering and other tuning > will fix this ? Or maybe the work mentioned getting qemu talking > directly to gluster will fix this? Have you tried setting performance.client-io-threads on if it makes any difference? As a side note I have to say that I have seen similar problems with RAID-5 systems even when using them as non-replicated iSCSI target. In my experience it's definetly not good for hosting VM images. -samuli
