Re: [Xen-users] dom0 console weirdness (reboot the only fix I've found)

[Mark Williamson <mark.williamson@xxxxxxxxxxxx>]

> hey ya'll! so I'll try to be as detailed as possible here. basically my xm
> console is getting corrupt somehow and the only fix I've found is a reboot.
> (lame)

Boo, that's not a great problem for you to be stuck with!  Have you tried a 
xend restart?  Is that not enough? :-/

> first, the details..
>
> [root@oak-dev-xen001 ~]# cat /etc/redhat-release
> CentOS release 5 (Final)
>
> [root@oak-dev-xen001 ~]# rpm -qa | grep -i xen | sort
> kernel-xen-2.6.18-8.1.15.el5.x86_64
> kernel-xen-2.6.18-8.el5.x86_64
> xen-3.0.3-25.0.4.el5.x86_64
> xen-libs-3.0.3-25.0.4.el5.x86_64
>
> [root@oak-dev-xen001 ~]# uname -a
> Linux oak-dev-xen001 2.6.18-8.1.15.el5xen #1 SMP Mon Oct 22 09:01:12 EDT
> 2007 x86_64 x86_64 x86_64 GNU/Linux
>
> (i've been able to reproduce this behaviour with both kernels listed above)
>
> ****
>
> I'm creating my (very basic) VM as follows:
>
> dd if=/dev/zero of="foo" bs=1048576 count=1 seek=20480 &> /dev/null
> mkfs.ext3 -qF foo
> mount -o loop foo /mnt/xen
>
> for i in console null zero mem port core full ram tty hda; do
>     MAKEDEV -a -d /mnt/xen/dev $i
> done

I think you need an xvc0 / hvc0 node in here for the Xen console device, 
depending on what kernel you're using.  For a RHEL kernel, I think it's xvc0.

> yum -t -e 0 -y --installroot=/mnt/xen install kernel
>
> ****
>
> my config file (left out networking foo to just get the thing to boot):
>
> [root@oak-dev-xen001 ~]# cat /etc/xen/configs/foo
> name = "foo"
> vif = [ '', 'bridge=xenbr0' ]
> disk = [ 'file:/data/xen/disks/foo,hda1,w' ]
> kernel = "/boot/vmlinuz-2.6.18-8.el5xen"
> ramdisk = "/boot/initrd-2.6.18-8.el5xen-no-scsi.img"
> memory= 1024
> root = "/dev/hda1 ro"
> extra = "4 enforcing=0"
> on_crash = 'destroy'
>
> ****

Have you tried including an explicit "console=xvc0" or somesuch in the extra= 
setting?

> NOW...the weird part.
>
> 'xm create -c /etc/xen/configs/foo' just hangs the console (no response)
> and eventually it exits with:
>   Error: destroyDevice() takes exactly 3 arguments (2 given)

That looks like a Xend bug, which is probably fouling up Xend for all your 
other domains as well :-(

> I tail the dom0 logs while starting up the domU and nothing stands out as
> erroring. once I corrupt the xen console via this method I can't boot ANY
> of my VMs -- even the ones I know are working. they all exit with the same
> error/blank console as above and require a reboot of the dom0 in order to
> boot successfully. once I reboot, I drop into the xen console without
> problem upon my next 'xm create' -- my VMs boot and everything works as it
> should.
>
> mind you this -only- occurs during new VM creation via the above 'dd/yum
> install' method above. prior to corrupting my console, I can copy existing
> VMs/start them up without incident - it's just the 'dd/yum install' method
> that screws things up and prevents me from starting up any of my VMs.

Interesting.  I guess it must be something about your setup there then - are 
you able to boot the "known good" domains with otherwise the same Xen config 
as the "known bad" setup (i.e. same kernel, initrd, etc)?

> my hypothesis? I think something confuses the dom0 during my VM creation,
> but I'm not sure where. my guess is the yum install.

You're correct about dom0 getting confused.  It's something exploding in Xend 
but I'm not 100% sure why.

Cheers,
Mark

-- 
Push Me Pull You - Distributed SCM tool (http://www.cl.cam.ac.uk/~maw48/pmpu/)

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users