- Users - libvirt List Archives

[libvirt-users] ISO refuses to let vm starts (and is not mentioned in config)
by Mauricio Tavares 07 Mar '14

07 Mar '14

Trying to start one of my vms, a centos one at that, but am getting the following message: [root@vmhost ~]# virsh start voip --paused error: Failed to start domain voip error: cannot open file '/var/tmp/FreePBX-5.211.65-3-x86_64-Full-1388073872.iso': No such file or directory [root@vmhost ~]# But, virsh dumpxml voip shows no info onto the .iso <disk type='file' device='disk'> <driver name='qemu' type='raw' cache='none' io='native'/> <source file='/dev/vmhost_vg0/voip'/> <target dev='vda' bus='virtio'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x04' function='0x0'/> </disk> <disk type='file' device='cdrom'> <driver name='qemu' type='raw'/> <target dev='hda' bus='ide'/> <readonly/> <address type='drive' controller='0' bus='0' target='0' unit='0'/> </disk> <disk type='file' device='cdrom'> <driver name='qemu' type='raw'/> <target dev='hdb' bus='ide'/> <readonly/> <address type='drive' controller='0' bus='0' target='0' unit='1'/> </disk> Log file is not particularly helpful: 2014-03-04 22:57:46.121+0000: shutting down 2014-03-05 23:42:18.528+0000: shutting down 2014-03-05 23:47:07.375+0000: shutting down 2014-03-06 00:56:32.908+0000: shutting down 2014-03-06 01:11:32.582+0000: shutting down So, the question is how did this iso got involved in this specific vm? I can start my other vms without them even knowing about this .iso.

1 2

[libvirt-users] Attaching a hostdev on startup
by Mat Arge 07 Mar '14

07 Mar '14

Hello! I have written a script which attaches some USB tokens as hostdev devices to (qemu) VMs managed by libvirt. This works fine as long as I call the script manually. But what I really want to do, is run this scrip from a startup hook, so that whenever a Vm is started, it looks if there are appopriate tokens available for attachment. But when the script is run from the hook, it (along with the whole libvirtd) freezes. That doesn't really come as a surprise, since http://libvirt.org/hooks.html explicitely states A hook script must not call back into libvirt, as the libvirt daemon is already waiting for the script to exit. A deadlock is likely to occur. How can I get around this problem? cheers Mat

1 0

[libvirt-users] Providing fallback DNS servers via DHCP
by Florian Rüchel 07 Mar '14

07 Mar '14

Hello everyone, TL;DR: Is ist possible to provide a secondary DNS (e.g. 8.8.8.8) via DHCP in addition to the gateway (in my case 10.0.1.1)? I use libvirt to manage all my qemu-based VMs. I have set up a local NAT-network which includes all my VMs (and some physical servers integrated via VPN). Everything works fine, but ocassionally, I have to restart the network (when I add instances for example). In that case, my VMs loose their local network connection until I restart them. However, some instances can not be restarted at will and so a change in network configuration means they loose their local connectivity. However, they still have a second network device that has an external static IP. This IP can still connect and be connected to, so it is feasible to loose the connection on the local network. Unfortunately though, libvirt (through dnsmasq's DHCP) broadcasts itself as the only DNS. This results in name resolution going down. If I add e.g. 8.8.8.8 to resolv.conf, I can resolve names again (except local ones, of course). So it would be nice if the DHCP would directly supply additional DNS servers. For dnsmasq, I would add to the configuration something like dhcp-option=6,0.0.0.0,8.8.8.8 However, looking through the archive, I found that you guys prefer to keep it implementation agnostic and thus you do not provide the option to set this directly (since this is dnsmasq-only). So I was wondering if there is any way to achieve this with libvirt? Is there already an implemented way I could not find? Or is there maybe a *generic* option to provide some kind of extra-configuration specific to the implementation? Or is my only solution to make sure the clients manually add a second DNS by themselves? Regards, Florian

1 0

[libvirt-users] Python-libvirt in a virtual environment.
by Sijo Jose 07 Mar '14

07 Mar '14

Hi, Could you guide me to install libvirt in a virtual environment..? OR How python-libvirt could be used within a virtual environment after installing it, in the system. Regards Sijo

3 4

[libvirt-users] how to disable vnet_hdr through libvirt
by Jianfeng Tang 06 Mar '14

06 Mar '14

Hi, I am using libvirt and virsh to create my VM on KVM. It works fine. However, I need to turn off vnet_hdr for virtio interface. Does anyone know how to do it through libvirt? Thanks.

3 5

[libvirt-users] Migration failure after libvirt upgradation
by Shweta Salaria 06 Mar '14

06 Mar '14

Team, I'm trying to migrate a domain offline(shutdown mode) using libvirt-python API 'migrateToURI' but it gives me an error : libvirt: QEMU Driver error : Requested operation is not valid: domain is not running When I tried the same API for live migration(running and suspend mode), it gives me an error: libvirt: QEMU Driver error : Unsafe migration: Migration may lead to data corruption if disks use cache != none Below are the versions of the virtualization softwares I am using currently : Hypervisor : KVM qemu : 1.0 qemu-kvm : 1.0 Libvirt : 1.2.1 Earlier, I had performed offline and live migration *successfully* with these versions: Hypervisor : KVM qemu : 1.0 qemu-kvm : 1.0 libvirt : 0.9.8 Also, I would like to know whether it's possible to migrate a domain having snapshots? Kindly guide me through this. Thanks in advance Regards, Shweta

1 0

[libvirt-users] How to enable mount loop device in LXC using libvirt LXC driver
by Du Jun 06 Mar '14

06 Mar '14

When I try to mount loop device in a linux container, I got the following error message: mount: Could not find any loop device. Maybe this kernel does not know about the loop device? (If so, recompile or `modprobe loop'.) I wonder how to enable mount loop device in LXC using libvirt LXC driver? -- Thanks! Du Jun

2 1

[libvirt-users] Fail to convert LXC container configuration into a domain XML fragment
by Du Jun 06 Mar '14

06 Mar '14

Hi, all, I used lxc-tools to create a linux container. I am trying to transfrom LXC container configuration into a domain XML fragment using the following command: *$ virsh -c lxc:/// domxml-from-native lxc /var/lib/lxc/my_conatiner/config* However, I get the error message, error: this function is not supported by the connection driver: virConnectDomainXMLFromNative It seems that *virConnectDomainXMLFromNative *is not supported in libvrit LXC driver. However, I find the command in the official documentation of libvirt. Besides, my libvirt's version is 1.2.2(the latest version). I wonder whether I can convert LXC container configuration into a domain XML fragment using *domxml-from-native*? If I can't, how cant I translate it? PS: I care about the rules in my LXC container configuration file, such as: *lxc.aa_profile = unconfined* *lxc.cgroup.devices.deny = a* *# Allow any mknod (but not using the node)* *lxc.cgroup.devices.allow = c *:* m* *lxc.cgroup.devices.allow = b *:* m* *lxc.cgroup.devices.allow = b 7:* rwm* I just don't know how to express the rules above using xml format. Please help! Thanks! -- Best Regards, Du Jun

2 1

[libvirt-users] Simple Networkfilter not working as expected.
by Matthias Babisch 06 Mar '14

06 Mar '14

Hello People. I have produced a very simple networkfilter that is not work as I would expect it. Perhaps one of you knows what I did wrong? I made this little filter: <filter name='my-test-no-ip-spoofing' priority='-700'> <rule action='drop' direction='out' priority='-999'> <all match='no' srcipaddr='$IP'/> </rule> </filter> I could attach it directly to a VM (and defined an IP-Adress in the network-interface there). Then it produced iptables rules that look like this: Chain FI-vnetnn (1 references) pkts bytes target prot opt in out source destination 0 0 DROP all -- * * ! IP 0.0.0.0/0 (This is the rule governing the input via the virtual device into the bridge, is as expected.) Chain HI-vnetnn (1 references) pkts bytes target prot opt in out source destination 0 0 DROP all -- * * ! IP 0.0.0.0/0 (This is the rule governing the input to the host, i would expect this too.) Chain FO-vnetnn (1 references) pkts bytes target prot opt in out source destination 0 0 DROP all -- * * 0.0.0.0/0 ! IP This is the rule governing the output via the virtual-device from the bridge. (i.e. Packets coming from the network.) I specifically asked to filter outgoing traffic. This one I don't unterstand. Perhaps somebody knows a hint? On the other hand this filter works as expected, no rule on "FO-vnetnn": <filter name='my-no-mac-spoofing' priority='-800'> <rule action='drop' direction='out'> <all match='no' srcmacaddr='$MAC'/> </rule> </filter> I used libvirt with qemu on Ubuntu 13.10. (Version 1.1.1-0ubuntu8.5) I am grateful for any helpful comments. Sincerely Matthias Babisch IT/Organisation *b+m Informatik AG* Rotenhofer Weg 20 24109 Melsdorf T +49 4340/404-1444 F +49 4340/404-111 M +49 160/8866426 matthias.babisch(a)bmiag.de Aktuelle Informationen unter www.bmiag.de <%5C%22http://www.bmiag.de%5C%22> Die b+m Informatik AG ist ein Unternehmen der Allgeier Gruppe <%5C%22http://www.allgeier-holding.de%5C%22> Vorsitzender des Aufsichtsrates: Dr. Marcus Goedsche Vorstand: Dipl-Ing. Frank Mielke Amtsgericht Kiel, HRB 5526

1 0

[libvirt-users] cgroup for VM - does it work properly?
by Martin Pavlásek 06 Mar '14

06 Mar '14

Hi I tried to restrict usage of some running VM by cpu.shares (i.e. set to 10 from original 1024) on loaded system and it seem doesn't work as I expected... all running processes has same CPU usage (by htop) :-/ Does anyone has same experience? Fedora 19, libvirt-1.0.5.9-1.fc19.x86_64 Thanks a lot Martin -- Martin Pavlásek <mpavlase(a)redhat.com> OpenStack QA Associate/Red Hat Czech/BRQ irc: mpavlase

4 5