Thanks for your help.
I have another question. After copying existing configuration file, If I
just change the name, UUID, and MAC, it works. However If I change the
vcpu in configuration file to a different number, for example, from 1 to
2, the new created domain will have some problems. When I was trying to
migrate this domain from one physical host to another one, the migration
never ends !!
Does this mean that creating new guests by copying existing configuration
file is only suitable to configure same-resource-allocation guests? If
vcpu of existing domain is 1, then the new domain has also to be 1??
Thanks for your time.
I want to configure a new domainU from existing domain. Can I just copy
the image file (/var/lib/xen/images/box) and also copy the configuration
file of existing domain?
The format of configuration file is like this:
# Automatically generated xen config file
name = "xen2"
memory = "1000"
disk = [ 'tap:aio:/var/lib/xen/images/box2,xvda,w', ]
vif = [ 'mac=xx:xx:xx:xx:xx:xx, bridge=xenbr1', ]
uuid = "9297c728-efe7-025f-3b35-02f92474aa6e"
on_reboot = 'restart'
on_crash = 'restart'
I will change the name to a new name, set vcpu to 4. But I don't know how
to get a unique mac address and uuid. Anybody know?
And except for create the image file and configuration file, what else do
i need to do?
PS: I use the Fedora XEN.
Thanks for your time.
I running a Fedora 7/i386 (hvm) guest on a F7 x86_64 host.
The guest was created by using virt-manager so it uses the new Xen 3.1 config
system. The installation worked fine but now the network card is missing in the
DomU. Even lspci in DomU can't detect the nic although it is configured in the
Dom0 (virsh dumpxml f7_32_build):
Interestingly, this problem seems to be related to libvirt or the new config
system... I created an old-style configuration file for the domain using the
uuid/mac address from virsh dumpxml and started the guest with 'xm create' =>
the network card reappeared in DomU.
After using the old config, I deleted the old-style config file and used 'virsh
start'. => Network card is still present in DomU.
I think the problem was reported (with Windows guests) on the mailing list already:
Loosing eth devices in DomU was reported by others too
* https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=246755 (comment 3)
Any ideas how to fix this problems? Why does the nic disappear in DomU?
I was thinking about this more and have not come up with a solid solution. Yes the new box does have Xen 3.0 and the older xen was running FC4 and I will assume Xen 2.0.
What is the best way to migrate the FC4 xen server over to the new server running FC6?
I want to get the image on the new server and then do an upgrade to it so it is running FC6.
> On Thu, Jul 19, 2007 at 03:51:21PM -0400, Formoso, Travis wrote:
> Minor update:
> I moved the vmlinuz-2.6.12-1.1398_FC4xenU to /boot on the new server and then ran:
> xm create -c /xen/sec-int-dns/xmconfig and got this error:
> Error: (22, 'Invalid argument')
> I'm guesing you're trying a i386 kernel here ? FC4 is using the obsolete
> non-PAE kernels, while FC6 and later are all PAE. You can't mix & match
> a PAE and non-PAE kernels in host & guest.
This e-mail and any files transmitted with it are for the sole use of
Blue Slate Solutions and the intended recipient(s) and may contain
confidential and privileged information. If you are not the intended
recipient, please contact the sender by reply e-mail and destroy all
copies of the original message. Any unauthorized review, use,
disclosure, dissemination, forwarding, printing or copying of this email
or any action taken in reliance on this e-mail is strictly prohibited
and may be unlawful.
On Tue, Jul 24, 2007 at 02:05:28PM +0200, Tadeas Mengler wrote:
> Thanks for the reply at first.
> You can see it in htop utility. The load comes from kernelspace.
> The symptoms are simillar to those:
What's the process causing the load / cpu-usage?
> The speed of for example of coping data across network is very slow
> even on gigabit network (about 3 mbit/s), while domU is extremly slow
> and unresponsive (all of the tasks).
Is the network performance slow to begin with, or after some time? If after
some time, what causes it to get slow?
Is network slow from dom0 to the network?
Is network slow between dom0 and domU ?
What NIC do you have?
What software are you using to test the network performance?
Have you tried iperf? both tcp and udp-mode? Do you see packet loss with udp?
Do you have any errors in logs/dmesg? both dom0 and domU? how about in "xm
log" or "xm dmesg" ?
Do you have any custom settings or xen scheduling parameters (domain
Are you using (the default) bridged networking setup with xen? Or a custom
> The hw varies, mostly supermicro servers, various cpus (different
> intels and amds), various ram , mostly sw raid. The disks are sas, sata.
> The distribution is centos5 with recent updates and current kernel from
> the distribution.
Ok, it should be fine then. I don't see such problems with my hardware and
the same hypervisor/kernels.
> Dne Tue, 24 Jul 2007 14:19:55 +0300
> Pasi Kärkkäinen <pasik(a)iki.fi> napsal(a):
> > On Mon, Jul 23, 2007 at 02:58:42PM +0200, Tadeas Mengler wrote:
> > > Hello
> > >
> > > I have Xen related issues. I try to describe it as best as I can.
> > > While transmitting large amounts of data from domU the IO load
> > > increases greatly, thus making the domU unusable. The load is
> > > significant even on dom0. The performance of cpus and hard drives is
> > > not an issue. We have this problem on all servers with diferent hw
> > > configurations. Please do you know the reason of this behaviour, do
> > > you have any experience whith such problems?
> > >
> > I'm going to ask you almost the same questions as on centos-list..
> > Nope, no such experience with many xen hosts.
> > How do you measure this "IO load" ? What effects does it have?
> > What kind of hardware do you have (cpu/mem/nic/disk or raid
> > controller/disks)?
> > Are you running centos5? latest updates installed? or fedora? what
> > fedora version? what kernels?
> > Can you post any performance data? How does this "io load" make domU
> > unusable?
> > I'm sorry byt your problem description lacks a lot of important
> > information..
> > -- Pasi
> S pozdravem
> Tadeas Mengler
> Senior Linux System Administrator
> 483 367 866, 608 264 866
> MITON CZ, s.r.o. | www.miton.cz
> Jiraskova 9, 466 01 Jablonec nad Nisou
> Tel.: 483 367 800, fax: 483 710 884
> www.stahuj.cz | www.stable.cz | www.otoman.cz | www.vareni.cz
/ - \
I have Xen related issues. I try to describe it as best as I can.
While transmitting large amounts of data from domU the IO load
increases greatly, thus making the domU unusable. The load is
significant even on dom0. The performance of cpus and hard drives is
not an issue. We have this problem on all servers with diferent hw
configurations. Please do you know the reason of this behaviour, do you
have any experience whith such problems?
Linux System Administrator
483 367 866, 608 264 866
MITON CZ, s.r.o. | www.miton.cz
Jiraskova 9, 466 01 Jablonec nad Nisou
Tel.: 483 367 800, fax: 483 710 884
www.stahuj.cz | www.stable.cz | www.otoman.cz | www.vareni.cz
i'm trying to install a new xen virtual machine on FC7, using virt-manager.
the wizard finishes successfully and it start to create a VM, the installation tree has been placed on host drive and shared using NFS, however installing a guest's linux starts but it couldn't mount the installation tree with the error "that directory could not be mounted from the server"!?
system specifications are:
[root@localhost ~]# uname -r
[root@localhost ~]# rpm -qa \*xen*
mounting from the host to itself with NFS is OK with this command:
[root@localhost ~]# mount -t nfs 192.168.100.99:/root/Desktop/fc7/ /root/Desktop/test1/
where 192.168.100.99 is the host IP and /root/Desktop/fc7/ is the address of installation tree on it!
i don't know the problem is with NFS, permissions or any thing else!
if more information is needed please just notify me, any advise, comment or solution is helpful!
Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection.
-----BEGIN PGP SIGNED MESSAGE-----
I just saved a minimal FC7 installation to a local file, and am in
the process of uploading it to Amazon S3 with the intention of
creating an Amazon EC2 virtual instance. The way I formatted my file
was quite simple... its just one ext3 root directory (so no swap,
boot, LVM, etc... directories). Has anyone done this before? Were you
successful? Amazon's Developer Guide suggests using Yum to install
the operating system, but I thought that using the Virtual Machine
Manager would be a lot easier. Anyways, I am in the (long) process of
uploading a 2.5 GB image to S3, so I wanted to hear some feedback as
to whether or not this will work as smoothly as I hope before I waste
any more time. Thanks in advance!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
-----END PGP SIGNATURE-----
On 22/07/07, Thomas Antony <thomas(a)antony.eu> wrote:
> I have the same problem with F7 x86_64 and all the windows domU's. When
> i reboot the server, every windows domU does not have a nic anymore. A
> reboot of the windows os does not help. I must shutdown the windows os,
> remove and add again a nic with virt-manager and then there is a nic
thanks, I hadn't tried to remove and re-add a new NIC, before I do
that I will run device manager with the special
environment variable to see if windows still knows about the NIC, but
thinks it is no longer present (maybe due to some MAC address change,
though it is present if I use virsh dumpxml)
> The paravirtualized F7 domU's are workin well.
> I think this is happening since i updated libvirt to version 0.3.0.
Not sure if this is identical to
certainly seems that whenever dom0 is rebooted the Windows domU will
loose it's NIC.