RAID-1 and swap and md when an IDE disc fails

I have been trying to find some "chapter and verse" discussion on using the md driver to host a RAID-1 swap partition.

Naturally this configuration is quite easy to setup!

My concern is more about operation in the event of a total disk failure. I have now suffered several failed discs (IDE discs FWIW) and on each occasion the system (Ubuntu LTS 6.06.1 Server) has tanked!

I presume this is due to the kernel needing to swap in/out some "stuff" at the time and on not finding the expected physical disk to talk to it went slightly catatonic. The keyboard was still active (echoing keystrokes) so parts of the kernel were still alive, but perhaps overall is was choked waiting for the outcome of the failed disc IO's.

Is this a kernel issue per se or an md issue? Has any research been done on the response to catastrophic disc failures.

Regards, Frank.
.