On Mon, 2 Feb 2009, der Mouse wrote:
Well, don't lose sight of the fact that before you hit ^C it wasn't that dead, since sshd must have been getting some user time or you wouldn't've gotten echo. sshd msut even have gotten some time immediately after, too, to echo the ^C, so it took at least a little while to lock up.
Hrrm, I left the machine in locked up state after the last hang.. Now I got curious and just checked: the box was actually responding to ping this time -- until I tried to ssh -vvv and nmap it, to see if it responded at all (it didn't). And now it doesn't respond to ping, either.
Hrrm, I remember seeing a very similar problem a long time ago, where the box would gradually lock up process by process, as they attempted disk io, until eventually it was fully emulating a dead duck. IIRC that was one of my spectacular usb2 umass failures on amd64 though, but this looks at least superficially similar.
/ali