Re: [Xen-devel] Failed to run bootloader

[Sandi Romih <romihs.forums@xxxxxxxxx>]

On Mon, Apr 23, 2012 at 1:30 PM, Ian Campbell <Ian.Campbell@xxxxxxxxxx> wrote:

On Mon, 2012-04-23 at 12:29 +0100, Stefano Stabellini wrote:
> On Mon, 23 Apr 2012, Ian Campbell wrote:
> > On Sat, 2012-04-21 at 12:04 +0100, Sandi Romih wrote:
> > > [...]
> >
> > >  kernel = "/etc/xen/kernels/oi151a/unix"
> > >  ramdisk = "/etc/xen/kernels/oi151a/boot_archive"
> >
> > I forget how this works -- do you need to comment these out while using
> > pygrub or are these paths inside the ISO?
> > [...]
> > >  disk =
> > > [ 'file:/mnt/media/comp_files/os-iso/oi151a.iso,xvdc:cdrom,r' ,
> > > 'phy:/dev/dom0/oi_boot,xvda,w' ]
> > >  bootloader = "/usr/bin/pygrub"
> >
> > Sadly I don't think xl is currently capable of booting using pygrub from
> > a file:/// type device.
>
> > I expect this will be resolved by Stefano's "qdisk local attach" series
> > which was on the list last week.
>
> Actually, even without my patch series, xl should be able to boot a PV
> guest using pygrub if (and only if) the disk is a raw file.

Yes, I've just realised that while editing
http://wiki.xen.org/wiki/Debian_Guest_Installation_Using_Debian_Installer as part of the dos day -- it seems to work for me at least.

Perhaps the issue here is the use of kernel= and ramdisk= instead of
bootloader_args? (see the new CD install section of that Debian install
guide for an example of the sort of thing which works for me)

Ian.

>
>
> > In the meantime as a workaround you could try using losetup on the iso
> > and pointing xl at phy:/dev/loop$N.
>
> that should also work

Hello guys,

Thanks for the responses.

I have been doing some testing, to see if I could find out what could be causing the problem.

I have, once again, reinstalled the whole system as I wanted a clean system, and now I can get the system to create my openindiana VM (with the bootloader property in the cfg file), however it will not boot fully.

What I did discover is that the VM boot is halted by the vif option in my cfg file. If I comment it out, the vm boots fine, but of course I cant complete the install without the network IF present.

I tried creating a different networking architecture as laid out in the xen wiki's, but with no difference in the behavior.

Occasionally I would get errors in the VM regarding network IF not being correctly initialized (or something to that extent), which reinforces my suspicions regarding networking causing the problems

I had all this working in my initial setup, 3.1.9 kernel with xen 4.2 unstable 14785, now I am running kernel 3.1.10. The major differences are that before I used a file based disk image while now I am using LVM disk image (or partition), and that I did not patch xen with the Solaris pvgrub patch. My current disk definition is:

 disk = [ 'file:/mnt/media/comp_files/os-iso/oi151a.iso,xvdc:cdrom,r' , 'phy:/dev/xen-dom0/oi_boot,xvda,w' ]

Have I perhaps made an error here?

Hope that I have not rambled on too much and confused you.

Best regards

Sandi

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel