Xen users - Sep 2009 - Always losing the console output after "Scrubbing free RAM"

Hello.

I have continued problems with an initial Xen run.  After "Scrubbing
free RAM" the monitor goes blank (existing output is erased) and it
stays that way.  The machine doesn''t come up pingable, it just goes
possum.

Thinking the console output is redirected, I have literally tried all
variants and combinations of passing parameters to the Linux kernel
that I''m loading in there.  Everything related to console=, xencons=,
novidmodechange(?) and a few more.  Not only didn''t it solve the
problem, there was no change at all, so I think I am on an entirely
wrong path here.

Setup:
- pxelinux diskless boot
- Debian''s build of Xen 3.4 hypervisor
- Debian''s kernel vmlinuz-2.6.30-1-amd64 which explicitly says it has
  Xen support in (and matching initrd and all three of xen, kernel and
  initrd are loaded fine according to the output I see).
- also tried my own 2.6.30.5 which also has about the same Xen options on (no
  initrd).  Same results.

The machine with the kernel provided boots fine with the diskless
setup and local config when booted outside of Xen.

label xen
        kernel mboot.c32
        append xen-3.4-amd64 dom0_mem=262144 noreboot console=vga --- \
		vmlinuz-2.6.30-1-amd64 console=tty0 console=hvc0 --- \
		initrd.img-2.6.30-1-amd64
#       append xen-3.4-amd64 dom0_mem=262144 noreboot --- \
		vmlinuz-2.6.30-1-amd64 console=tty1 xencons=tty1 ---
		initrd.img-2.6.30-1-amd64

(backslashed line breaks only for mail, not in the file)

The hardware is a Dual-Core Opteron socket 939 in a Via K8T800 board
(Asus A8V-E) with some modern ATI graphics card.

Questions:

- let''s say I use an entirely unsuitable kernel (no Xen) support, what
  exactly would the symptoms be?

- is that hardware maybe unsupported? What are the symptoms if I have
  unsupported hardware? Are NVidia NForce chipsets better supported?

What''s the best way to proceed here? Should I build Xen and the Xen
Linux kernel from your sources? Should I try different hardware?
Serial console?

Thanks a lot
Martin
-- 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Martin Cracauer <cracauer@cons.org>   http://www.cons.org/cracauer/
FreeBSD - where you want to go, today.      http://www.freebsd.org/

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

Still losing the console output.  I now used the 2.6.18-xen that comes
from http://www.xen.org/products/xen_source.html instead of the Debian
kernel and I still have this problem, no change.

Could somebody give my questions below a quick pass? I''m really not
sure what I''m missing here.

Thanks
	Martin

Martin Cracauer wrote on Mon, Sep 07, 2009 at 10:27:26AM -0400:
> Hello.
> 
> I have continued problems with an initial Xen run.  After "Scrubbing
> free RAM" the monitor goes blank (existing output is erased) and it
> stays that way.  The machine doesn''t come up pingable, it just
goes
> possum.
> 
> Thinking the console output is redirected, I have literally tried all
> variants and combinations of passing parameters to the Linux kernel
> that I''m loading in there.  Everything related to console=,
xencons=,
> novidmodechange(?) and a few more.  Not only didn''t it solve the
> problem, there was no change at all, so I think I am on an entirely
> wrong path here.
> 
> Setup:
> - pxelinux diskless boot
> - Debian''s build of Xen 3.4 hypervisor
> - Debian''s kernel vmlinuz-2.6.30-1-amd64 which explicitly says it
has
>   Xen support in (and matching initrd and all three of xen, kernel and
>   initrd are loaded fine according to the output I see).
> - also tried my own 2.6.30.5 which also has about the same Xen options on
(no
>   initrd).  Same results.
> 
> The machine with the kernel provided boots fine with the diskless
> setup and local config when booted outside of Xen.
> 
> label xen
>         kernel mboot.c32
>         append xen-3.4-amd64 dom0_mem=262144 noreboot console=vga --- \
> 		vmlinuz-2.6.30-1-amd64 console=tty0 console=hvc0 --- \
> 		initrd.img-2.6.30-1-amd64
> #       append xen-3.4-amd64 dom0_mem=262144 noreboot --- \
> 		vmlinuz-2.6.30-1-amd64 console=tty1 xencons=tty1 ---
> 		initrd.img-2.6.30-1-amd64
> 
> (backslashed line breaks only for mail, not in the file)
> 
> The hardware is a Dual-Core Opteron socket 939 in a Via K8T800 board
> (Asus A8V-E) with some modern ATI graphics card.
> 
> Questions:
> 
> - let''s say I use an entirely unsuitable kernel (no Xen) support,
what
>   exactly would the symptoms be?
> 
> - is that hardware maybe unsupported? What are the symptoms if I have
>   unsupported hardware? Are NVidia NForce chipsets better supported?
> 
> What''s the best way to proceed here? Should I build Xen and the
Xen
> Linux kernel from your sources? Should I try different hardware?
> Serial console?
> 
> Thanks a lot
> Martin
> -- 
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> Martin Cracauer <cracauer@cons.org>   http://www.cons.org/cracauer/
> FreeBSD - where you want to go, today.      http://www.freebsd.org/
-- 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Martin Cracauer <cracauer@cons.org>   http://www.cons.org/cracauer/
FreeBSD - where you want to go, today.      http://www.freebsd.org/

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

On Mon, Sep 7, 2009 at 9:27 PM, Martin Cracauer<cracauer@cons.org>
wrote:
> Setup:
> - pxelinux diskless boot
> - Debian''s build of Xen 3.4 hypervisor
> - Debian''s kernel vmlinuz-2.6.30-1-amd64 which explicitly says it
has
>  Xen support in (and matching initrd and all three of xen, kernel and
>  initrd are loaded fine according to the output I see).
> - also tried my own 2.6.30.5 which also has about the same Xen options on
(no
>  initrd).  Same results.
One at a time :D

Were you able to get Xen to run succesfully with local boot? If not,
try it. At least that would help determine whether the problem is
Xen/Debian setup, or in mboot/pxe.

-- 
Fajar

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

Fajar A. Nugraha wrote on Fri, Sep 11, 2009 at 09:47:37AM +0700:
> On Mon, Sep 7, 2009 at 9:27 PM, Martin Cracauer<cracauer@cons.org>
wrote:
> > Setup:
> > - pxelinux diskless boot
> > - Debian''s build of Xen 3.4 hypervisor
> > - Debian''s kernel vmlinuz-2.6.30-1-amd64 which explicitly
says it has
> >  Xen support in (and matching initrd and all three of xen, kernel and
> >  initrd are loaded fine according to the output I see).
> > - also tried my own 2.6.30.5 which also has about the same Xen options
on (no
> >  initrd).  Same results.
> 
> One at a time :D
> 
> Were you able to get Xen to run succesfully with local boot? If not,
> try it. At least that would help determine whether the problem is
> Xen/Debian setup, or in mboot/pxe.
I prepared a regular disk and booted from grub, on the K8T800 chipset
machine and on a NForce machine.  Same result, the console disappears
after "scrubbing free RAM" and never comes back.  No change in
symptoms between pxelinux and grub boots.

I would really appreciate it if you could give these questions a try:

1) if I had a dum0 kernel that''s just unsuitable (code or config),
what would the symptoms of trying to boot it be?

2) who''s printing "scrubbing free ram", the hypervisor or
dom0?

2b) speaking of who''s printing what, at this point is there any sign
that dom0 even started doing some work?

Thanks
	Martin
-- 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Martin Cracauer <cracauer@cons.org>   http://www.cons.org/cracauer/

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

On Mon, Sep 14, 2009 at 7:54 AM, Martin Cracauer <cracauer@cons.org>
wrote:
> Fajar A. Nugraha wrote on Fri, Sep 11, 2009 at 09:47:37AM +0700:
>> Were you able to get Xen to run succesfully with local boot? If not,
>> try it. At least that would help determine whether the problem is
>> Xen/Debian setup, or in mboot/pxe.
>
> I prepared a regular disk and booted from grub, on the K8T800 chipset
> machine and on a NForce machine.  Same result, the console disappears
> after "scrubbing free RAM" and never comes back.  No change in
> symptoms between pxelinux and grub boots.

Well, this is why I said "One at a time". You''re trying to
create a
complicated setup without even knowing the basics. I suggest you try
using vendor-bundled Xen first to boot from local disk. Example:
RHEL/Centos5, Opensuse 11.1, Debian Etch.
>
> I would really appreciate it if you could give these questions a try:
Here''s what I got from a 64bit Opensuse 11.1 :
>
> 1) if I had a dum0 kernel that''s just unsuitable (code or config),
> what would the symptoms of trying to boot it be?
(XEN) *** LOADING DOMAIN 0 ***
(XEN) elf_xen_note_check: ERROR: Will only load images built for the
generic loader or Linux images
(XEN) ****************************************
(XEN) Panic on CPU 0:
(XEN) Could not set up DOM0 guest OS
(XEN) ****************************************
(XEN)
(XEN) Reboot in five seconds...

>
> 2) who''s printing "scrubbing free ram", the hypervisor
or dom0?
Should be hypervisor. But it only done so after it finds a suitable
kernel to load

(XEN) *** LOADING DOMAIN 0 ***
(XEN)  Xen  kernel: 64-bit, lsb, compat32
(XEN)  Dom0 kernel: 64-bit, lsb, paddr 0x200000 -> 0x7e20b8
(XEN) PHYSICAL MEMORY ARRANGEMENT:
(XEN)  Dom0 alloc.:   0000000078000000->000000007a000000 (465943 pages
to be allocated)
(XEN) VIRTUAL MEMORY ARRANGEMENT:
(XEN)  Loaded kernel: ffffffff80200000->ffffffff807e20b8
(XEN)  Init. ramdisk: ffffffff807e3000->ffffffff81339e00
(XEN)  Phys-Mach map: ffffffff8133a000->ffffffff816d80b8
(XEN)  Start info:    ffffffff816d9000->ffffffff816d94b4
(XEN)  Page tables:   ffffffff816da000->ffffffff816e9000
(XEN)  Boot stack:    ffffffff816e9000->ffffffff816ea000
(XEN)  TOTAL:         ffffffff80000000->ffffffff81800000
(XEN)  ENTRY ADDRESS: ffffffff80200000
(XEN) Dom0 has maximum 1 VCPUs
(XEN) Scrubbing Free RAM: .done.

>
> 2b) speaking of who''s printing what, at this point is there any
sign
> that dom0 even started doing some work?
Here''s the log continuation from (2a)

(XEN) Scrubbing Free RAM: .done.
(XEN) Xen trace buffers: disabled
(XEN) Std. Loglevel: Errors and warnings
(XEN) Guest Loglevel: Nothing (Rate-limited: Errors and warnings)
(XEN) Xen is relinquishing VGA console.
(XEN) *** Serial input -> DOM0 (type ''CTRL-a'' three times
to switch
input to Xen)
(XEN) Freed 132kB init memory.
Kernel alive
Kernel really alive
Initializing cgroup subsys cpuset
Initializing cgroup subsys cpu
Linux version 2.6.27.7-9-xen (geeko@buildhost) (gcc version 4.3.2
[gcc-4_3-branch revision 141291] (SUSE Linux) ) #1 SMP 2008-12-04
18:10:04 +0100
Command line: root=LABEL=ROOT splash=silent showopts console=tty1 console=xvc0

Lines from "Linux version" should be from dom0 kernel. If you
don''t
have that, probably somehow you don''t have the correct console
initialized. Probably incorrect paramater or some missing kernel
compile option. In my case, I setup xen to use serial port 1 as
console, and the kernel uses Xen console (xvc0) and vga (tty1). Here''s
the relevant grub.conf entry

title Xen dom0
    root (hd0,1)
    kernel /boot/xen.gz com1=115200,8n1 console=vga,com1 dom0_vcpus_pin
    module /boot/vmlinuz-2.6.27.7-9-xen root=LABEL=ROOT splash=silent
showopts console=tty1 console=xvc0
    module /boot/initrd-2.6.27.7-9-xen

Now it doesn''t always need to be THAT complicated. To find out what
parameters you need, it''s easier to let your distro setup the correct
parameters first by installing it''s version of Xen. After that, you
can modify it up to your needs.

-- 
Fajar

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

> > I prepared a regular disk and booted from grub, on the K8T800 chipset
> > machine and on a NForce machine.  Same result, the console disappears
> > after "scrubbing free RAM" and never comes back.  No change
in
> > symptoms between pxelinux and grub boots.
[no more problems, I just want to put this into the mail archive in
case people come across the same problem]

I''m all up and running fine having restarted from a fresh Debian 4.0
Etch.

Trying to isolate the failure, replacing Etch parts (hypervisor,
kernel) with Debian 5.02 Lenny parts fails.  Using the 3.4 Hypervisor
from Lenny comes up with Dom0 but xend fails and no DomUs come up.
Using the 3.4 Hypervisor from lenny and the lenny 2.6.30*xen* kernel
fails the dom0 boot.  diskless/PXE seem not to have been involved in
this problem, and my hardware seems fine, it''s just the new Debian
parts failing (possibly in combination with my hardware).  I have also
seen references that the new gear might have amd64 problems.

Anyway, thanks for your help and I''m up and running now.

Parts used now:
- Debian 4.0r8-amd64 install disk
- xen-hypervisor-3.0.3-1-amd64
- linux-image-2.6.18-6-xen-amd64
- linux-modules-2.6.18-6-xen-amd64
- xen-tools 2.8 
- xen-utils-3.0.3-1
- xen-ioemu-3.0.3-1

Hardware is a Asus A8V-E (K8T800 chipset), P-ATA disk, sky2 onboard
GbE, ATI 9200 video card.

Martin
-- 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Martin Cracauer <cracauer@cons.org>   http://www.cons.org/cracauer/

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

On Tue, Sep 22, 2009 at 10:40 PM, Martin Cracauer <cracauer@cons.org>
wrote:
> [no more problems, I just want to put this into the mail archive in
> case people come across the same problem]
Glad to hear you got it working.
>
> I''m all up and running fine having restarted from a fresh Debian
4.0
> Etch.
>
> Trying to isolate the failure, replacing Etch parts (hypervisor,
> kernel) with Debian 5.02 Lenny parts fails.  Using the 3.4 Hypervisor
> from Lenny comes up with Dom0 but xend fails and no DomUs come up.
Isn''t Xen 3.4 and 2.6.30 kernel only available in sid, not Lenny?
Anyway, the part where xend failed is probably due to mismatch between
hypervisor - userland tools. Although you can (usually) mix-and-match
dom0 kernel (at least on linux anyway), hypervisor and userland (xend,
xm, etc.) version must match.
> Parts used now:
> - Debian 4.0r8-amd64 install disk
> - xen-hypervisor-3.0.3-1-amd64
That one''s kinda old, so if you have some time and need newer versions
you might want to try compiling xen 3.4.1 (at least the hypervisor and
userland part) yourself.
> - linux-image-2.6.18-6-xen-amd64
> - linux-modules-2.6.18-6-xen-amd64
Although 2.6.18 is old, depending on your needs this might suit you
better. So if it works for your hardware I suggest stick with 2.6.18
(either form Etch, or compile yourself from xen.org''s kernel).

-- 
Fajar

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