a) Notice one NFS Mount from Fileserver ain’t working any more, and the idiotic Debian VM of course locked up instead of doing something graceful where it lost a non-root disk. Instead of listening to music on your morning coffee you’re now looking at a full-scale data loss?
b) Notice the following kernel message on the host
raid10_make_request bug: can't convert block across chunks or bigger than 64K
Find out you’re hit by https://bugzilla.redhat.com/show_bug.cgi?id=224077
Make it “work again” by using tap:aio and remain a little puzzled. Doesn’t Xen always break IOs into 64K chunks?
3) Stop wondering. Hit ebay and order a Dell Perc5 or Perc6.