Xen users - Oct 2006 - "Error: an integer is required", starting hvm machines during boot, kernel errors

Hello,

I often get the message "Error: an integer is required" when starting
an
VM, the VM starts anyways. At the next start of the VM, the error
message don''t appears... I don''t change anything on the
config. Its very
curious :o What can generate this error?  The config file from the VM
is attached.

I try to start automatically an HVM during boot.

If I define on_crash="destroy" in the VM configfile, the machine
won''t
boot correctly because the first try to boot it fails.
I get the following error message in qemulog:
"Failed allocation for dom 5: 1025 pages order 0 addr_bits 0".
I attached the xend logfile and the qemu log.
If I define on_crash="reboot" in the configfile, the VM successfull
starts at bootup with the state "----".
Why the machine successfull starts after 1 retry when
on_crash="reboot"
is set?
What means the "Failed allocation..." error message?
What means the state "----"?

Also now and then the machines boots autonomously without any cognizable
reason. It seems that this only appears when an HVM was automatically
started during bootup. There aren''t any erorrs in the logfiles...
If i start the same VM after booting manually with "xm create" it runs
without problems...
I will let run memtest86 soon, so I can be really sure that defect RAM
isn''t the cause for this behaviour

Sometimes when I close an SDL HVM window I get the kernel error attached
to this mail... What can be the reason for this?

In the HVM I talk about is a Linux 2.6.x  running... Please don''t
recommend me to let it run as PVM :) I have reasons to run it as HVM and
don''t won''t discuss about it... Linux as HVM should work
afaik.

CPU:  Intel(R) Pentium(R) D CPU 3.00GHz
Distribution: Debian
Kernel: Linux version 2.6.16-2-xen-686 (Debian 2.6.16-16bpo1)
xen-hypervisor-3.0-i386        3.0.2+hg9697-0bpo1
xen-ioemu-3.0                  3.0.2+hg9697-0bpo1


thanks for help

greetings

Fabian







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

On 05.10.2006 12:50, Fabian Holler wrote:
> I will let run memtest86 soon, so I can be really sure that defect RAM
> isn''t the cause for this behaviour
memtest86 runs 23 hours and found no Errors.
It seems like it is a Software Problem...


greetings

Fabian






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

Hi Fabian,

I would try running the test for longer.  I once had a faulty memory DIMM that I
was confident was faulty as system instabilities
followed the memory DIMM across multiple computers.  On one system it took over
2 days for memtest86 to detect an error!  Just to
check, I swapped out the faulty DIMM for a good one and ran memtest86 for a week
with no errors.  The DIMM was definitely faulty,
and it did make that system unstable, but it took much more than 23 hours for a
failure to occur using memtest86.

BR,

Roger
> -----Original Message-----
> From: xen-users-bounces@lists.xensource.com
[mailto:xen-users-bounces@lists.xensource.com] On
> Behalf Of Fabian Holler
> Sent: 06 October 2006 11:29
> To: xen-users@lists.xensource.com
> Subject: Re: [Xen-users] "Error: an integer is required",
starting hvm machinesduring boot, kernel
> errors
> 
> On 05.10.2006 12:50, Fabian Holler wrote:
> > I will let run memtest86 soon, so I can be really sure that defect RAM
> > isn''t the cause for this behaviour
> 
> memtest86 runs 23 hours and found no Errors.
> It seems like it is a Software Problem...
> 
> 
> greetings
> 
> Fabian
> 
> 
> 


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