From: Daniel Heemann <daniel.heemann@gm...> - 2003-01-23 15:13:20
I'm currently running 6 uml instances on an old k6-2 350, 192mb ram, enou=
swap, for testing purposes, respectively they serve some services for our=
living community (but as the tech specs should show it's not a production=
environment ;) ).
Host and uml instances both run Debian Woody, the host runs kernel 2.4.20=
skas support) and the uml instances 2.4.19.um46. The host kernel is patch=
for xfs. The uml kernel contains some patches for the reiserfs and=20
CONFIG_BLK_DEV_UBD_SYNC is enabled.
Almost everything works fine (ok, sometimes it's a little bit slow, when =
host has to swap but that's no problem for me) but sometimes an uml insta=
hangs with 'I'm tracing myself and I can't get out'.
It seems that the problem only occurs under heavy load of the uml instanc=
itself, mostly when running daily cronjobs or one uml only runs a syslogd=
the whole net, here you can see that if the uml hangs it hangs while runn=
logcheck (which will eat as much cpu as possible).
At least the problem can't be reproduced, it perhaps occurs 1x/2 weeks or=
similar, mostly the syslog/logcheck-uml hangs while running logcheck.
Perhaps it sounds like a hardware defect because it only hangs under heav=
load but I don' think so as the host itself runs stable w/o any problem a=
under heavy load of other processes.
You'll find a gdb output under the following address (but this time it se=
as the uml has not been under heavy load while hanging):
It has been created as described here:
"Those who would give up essential liberty, to purchase a little temporar=
safety, deserve neither liberty nor safety." - Benjamin Franklin