From: Ian Chilton <mailinglist@ic...> - 2002-11-09 17:13:09
I have been trying to setup my new server running UML's and twice
I have been compiling inside UML (different ones) and it has hung.
When you look at the uml's console, you see:
"I'm tracing myself and I can't get out".
uml_mconsole then fails to kill it - it just sits there.
Last night when it happened the 1st time I killed all the processes off
and restarted the uml and when I tried to continue the compile it was
all screwed so I deleted it and tried rextracting it, it spewed out with
loads of I/O errors. Later that night I was trying to work out why named
on the host kept dying when I realised everything was totall screwed - I
was getting Input/Output error on most things and couldn't even
shutdown/reboot. I had to hit the reset on the box.
Today mysql (which caused the above problem) compiled fine. I've since
compiled several things and it all seemed to be working fine until a
postfix compile did the same thing. I've not yet killed the processes
yet so I dont know whether the host will survive.
Host is running 2.4.20-pre11-fairsched:
uml@...:~]$ uname -a
Linux buzz 2.4.20-pre11 #2 SMP Sun Nov 3 17:35:04 GMT 2002 i686 unknown
UML's are running 2.4.19-24um.
I'm going to try updating the host and uml kernels to the latest now
(-rc1 and 28um last time I checked).
If it happens again after this, I will compile a UML kernel with
debugging and see if I can get a backtrace.
Has anyone else had similar problems?
From: David Coulson <david@da...> - 2002-11-10 03:29:12
Ian Chilton wrote:
> When you look at the uml's console, you see:
> "I'm tracing myself and I can't get out".
As I said on IRC, boot all your UMLs up with 'debug' and then wait for
one to panic. Then just bt it and mail the output here.
> uml_mconsole then fails to kill it - it just sits there.
Yup, a panic would do that.
David Coulson http://davidcoulson.net/