Strange LTSP phenomenon: Clients sessions freeze

[RalfGesellensetter <rgx@gmx.de>]

Hi there, 

as you might remember, I replaced fam by gamin, installed and,
and also killer.

Not much later, there were 2 incidences, one at 2.18 pm, next at 3.18 pm 
(one hour apart). When I read that "killer" is run by a cronjob every 
hour, I ascribed this phenomenon to "killer", but it doesn't make sense:

All running sessions were freezing.
Checking on what was happening through ssh (being logged in), I could 
tell
- eth1 was still up and ok.
- htop didn't show any CPU blockings
- dmesg told me "firefox-bin segfaults", and also some verbose 
  information about CPU 0 and 1 detached (or sim.) I had never seen    
  before (I could provide further details), but nothing really
  significant.

It was just as if all thin clients nics had stopped working at once.
It was still possible to boot further thin clients or reboot the frozen 
ones. Any idea what to look for?

Thank you
Regards
Ralf