Some more thoughts on this. In theory ( I have never done it myself or heard of
it being done), you should be able to kill a stuck process without rebooting if
you write a kernel module that will modify the kernel structures cleverly enough
to convince the kernel to permanently fail on that I/O. You would REALLY have to
know what you are doing, though. Maybe Byron could comment on this tomorrow night...
--
Sasha Pachev
AskSasha Linux Consulting
http://www.asksasha.com
.===================================.
| This has been a P.L.U.G. mailing. |
| Don't Fear the Penguin. |
| IRC: #utah at irc.freenode.net |
`==================================='