Am Dienstag, 17. Januar 2006 23:28 schrieb prosolutions@xxxxxxx:
> So wrote prosolutions@xxxxxxx on Tuesday, 17 January 2006:
[...]
> > Thank you. I wasn't clear about that. I apt-get installed
> > linux-xen0-2.6 but didn't realize after this the need to manually
> > install the .deb in /usr/src. I was expecting a debconf dialog with
> > information and neglected to see the output message.
no, just some simple echos to std.out in the postinst scripts for now... I
might change to using templates and debconf in future.
linux-xen0-2.6 is just a kind of container/meta-package. It helps me compiling
the kernel for each debian distro and arch without having to do it manually.
It also gives me the possibility to divert the kernel from the hypervisor &
userspace tools. I don't need to provide a new kernel when I build a new xen
package. Only on new upstream versions of xen a new kernel is really needed.
> > I will keep
> > plugging away at this and let you know my results. By the way, I am
> > installing this on an HP Proliant DL380 with a Compaq 5i SCSI RAID
> > controller and if we get this working it will be the first time I've
> > seen Xen run on this platform.
>
> Sorry to report that the booting of this version of Xen also failed on
> the Proliant DL380 running Debian/stable.
hmmm, I can't support that kind of problems myself, because that is far away
from "packaging-problems"... but maybe some other here can.
but was always may be worth investigating is the bios setting of your server.
try to disable every not really needed feature (like usb, onboard sound,
etc...) maybe some bios switch might help you to boot the kernel.
also using the noacpi boot-parameter might help. don't know...
ah, and if you are using the pae version, try the normal version instead.
might work better... is also helps to exclude pae as your problem.
> I cannot give you much
> information as the bootup was very fast and what initially looked like a
> normal bootup quickly went to a blank screen and the machine promptly
> rebooted.
oh, there is way to get a bit more output... is the dom0 kernel
hanging/rebooting or the xen hypervisor?
I never saw the hypervisor crash, but if something is wrong with your dom0,
then you should append "noreboot" to menu.list line regarding the kernel.
for example:
...
module /boot/xen-linux-2.6.12.6-xen root=/dev/hda7 ro noreboot
...
then you can at least try to see when the kernel crash... that could maybe
help to see where the problem is.
> I think the only way I am going to get Xen to run on this hardware is to
> build my own kernel -
That would only help, if you find the part that is causing the trouble and
when you don't need this part, so you can safely disable it.
I have exactly this kernel running on a dual xeon smp machine with 3ware sata
controller and on a dual xeon machine with a adaptec scsi controller. both
are orig. intel boards, but not from a distributor like dell, hp or so...
they are both running without any issues.
> but all my experience with Xen so far is that it
> is essentially so brittle as to be virtually unusable.
to help you, if you want to try it with your own kernel, take a look at:
http://www.debian-administration.org/articles/320 (again thx to Steve Kemp)
It's written for sid(unstable) but also working 1:1 for sarge...
> That is sad,
> because I have seen a lot of positive press about it and really was keen
> to start testing it for use in a production environment.
it's really great... I have running about 12 domains on a dual xeon with 3GB
ram without any performance issues... with a bit more ram I think about 20
(any maybe more) production domainU are realistic. If I see that correctly on
our box, the first bottleneck will be the disk drives (but at the moment it's
ok), because there is quite a lot of disk usage from the domainUs in our
setup. cpu usage is quite moderate for 12 productions systems and would be
enough for even more domains.
> Thanks for your help.
btw... so that others can better help you... do you use xen3 in 32bit mode or
xen3 32bit + pae or even xen3 in 64bit mode? that could be quite important.
ah, before I forget to mention, you are running xen3-testing with changeset
8269. that could be important for other to know and for further debugging.
sorry... I can't do anything more, but I am not a developer or a kernel
guru :)
--Ralph
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
Home