- Users - libvirt List Archives

[libvirt-users] about guest live migration configuration changes persistence ..
by Jean-Pierre Ribeauville 03 Sep '15

03 Sep '15

Hi, Is a change , by edit it via virsh edit , in the "local" .xml configuration file (i.e. adding a channel device) persistent when live migrating the guest on another host ? Thx. J.P. Ribeauville P: +33.(0).1.47.17.27.87 Puteaux 3 Etage 5 Bureau 4 jpribeauville(a)axway.com<mailto:jpribeauville@axway.com> http://www.axway.com<http://www.axway.com/> P Pensez à l'environnement avant d'imprimer.

3 5

[libvirt-users] New libvirt guide
by David Ashley 02 Sep '15

02 Sep '15

All - Let me introduce myself. I am W. David Ashley, the primary author of the "Libvirt Application Development Guide Using Python" which will soon be published on the libvirt.org web site. I hope all of you will enjoy the new guide when it becomes available. I have multiple decades of experience in writing documentation and training guides and I have about eight years of experience using libvirt and qemu/kvm virtual machines. Most of that experience was in creating and maintaining VM on-demand systems. In the meantime, I am planning a new guide and I would appreciate some feedback from the libvirt user community concerning the potential usefulness and contents of the new guide. The proposed title of the guide will be "Automating Virtual Machines". The current (very) rough outline is: Introduction Intro to virtual machines Installing virtual machines Using virtual machines Using Python to access VMs A sample problem Problem statement Solution requirements Using VMs to solve the problem New problems introduced by using VMs How to programmatically access VMs Intro to the python libvirt module How to connect to a VM How to access and control a guest domain Information the python libvirt module can not provide Storing information about VMs Deciding what information should be stored Using simple text files Using a simple database Securing your information How to set up a VM on demand environment Discovering a VMs ip address Using cron to start processes Starting up a VM Invoking a program on the VM Using SSH to access a VM Shutting down a VM Alternatives to starting/stopping a VM Using the libvirt guest agent Installing the libvirt guest agent Red Hat, Fedora, CentOS openSuse, SuSE Ubuntu, Debian, Mint Using (querying) the libvirt guest agent Logging VM activities Host activities VM activities Securing your VMs Host security VM security Any feedback/suggestions you have will be appreciated and I assure you they will be given serious consideration. At this point, nothing has been written except this rough outline so this is your chance to help form the contents or even make suggestions for a completely different guide. Feel free to post back to this list or send me private email. W. David Ashley w.david.ashley(a)gmail.com

3 2

[libvirt-users] VM locking
by Prof. Dr. Michael Schefczyk 01 Sep '15

01 Sep '15

Dear All, I am trying to use VM (disk) locking on a two node Centos 7 KVM cluster. Unfortunately, I am not successful. Using virtlockd (https://libvirt.org/locking-lockd.html) I get each host to write the zero length file with a hashed filename to the shared folder specified. Regardless of which host I start a VM (domain) on, they do produce the identical filename per VM. What does not work, however, is to prevent the second host to start the VM already running on the first VM. Using sanlock (https://libvirt.org/locking-sanlock.html) no domains do start at all. However, the file "__LIBVIRT__DISKS__" is also not written to the shared folder. Hence, nothing works. Can someone please point me to the right direction? My system is current Centos 7 using a gluster storage setup like suggested for oVirt (https://community.redhat.com/blog/2014/10/up-and-running-with-ovirt-3-5/) based on the oVirt 3.5 repo but without an engine. I do this, because I want to retain VM files in qcow2 format for live backups and with human readable names. What does work then is, e.g., live migration. A locking mechanism to prevent starting a VM twice would be good. Please note that this configuration - both according to redhat and to my own trial and error - requires Lock=False in /etc/nfsmount.conf. Is there a connection with my findings? My issues occur regardless of the files being in NFS or Gluster folders. KVM must load the Gluster storage indirectly via NFS rather than straight via Gluster, as Gluster storage does not seem to fully work, at least in virt-manager. My software versions are: libvirt 1.2.8-16.el7_1.3.x86_64 qemu-kvm ev-2.1.2-23.el7_1.3.1.x86_64 Regards, Michael

2 2

[libvirt-users] Live migration & storage copy broken since 1.2.17
by Antoine Millet 31 Aug '15

31 Aug '15

Hi, It seems that live migration using storage copy is broken since libvirt 1.2.17. Here is the command line used to do the migration using virsh: virsh migrate --live --p2p --persistent --undefinesource --copy -storage-all d2b545d3-db32-48d3-b7fa-f62ff3a7fa18 qemu+tcp://dest/system XML dump of my storage: <pool type='logical'> <name>local</name> <uuid>276bda97-d6c2-4681-bc3f-0c8c221bd1b1</uuid> <capacity unit='bytes'>1024207093760</capacity> <allocation unit='bytes'>4294967296</allocation> <available unit='bytes'>1019912126464</available> <source> <name>hkvm</name> <format type='lvm2'/> </source> <target> <path>/dev/hkvm</path> </target> </pool> When it fails, I get the following error: error: Storage volume not found: no storage vol with matching name '402aea08-26aa-45b6-a46e-a3b02137ff26' Where 402aea08-26aa-45b6-a46e-a3b02137ff26 is the volume bound to my VM: <disk type='volume' device='disk'> <driver name='qemu' type='raw' cache='none'/> <source pool='local' volume='402aea08-26aa-45b6-a46e -a3b02137ff26'/> <backingStore/> <target dev='vda' bus='virtio'/> <alias name='virtio-disk0'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x04' function='0x0'/> </disk> Here is the test result (version is for both source and destination): - 1.2.15 -> ok - 1.2.16 -> ok - 1.2.17 -> fails - 1.2.18 -> fails Is it a bug? Do I miss something? Thanks, Antoine

1 1

[libvirt-users] CPU feature 'svm' not reported inside guest without 'host-passthrough'
by Sladjan Ristic 30 Aug '15

30 Aug '15

Hi, I am confused by the behaviour of the CPU description in my VM. I have a host with an AMD CPU with SVM feature and I want to try nested virtualization on a Fedora22 guest. The host is Fedora22 as well. libvirtd (libvirt) 1.2.13.1 kernel 4.1.6-200.fc22.x86_64 1. I tried 'custom' mode with model qemu64 with required feature 'SVM' but '/proc/cpuinfo' in the guest doesn't show SVM. 2. However, if I forbid the SVM feature, the guest can't be created and libvirt complains that the host CPU has SVM. 3. The only way to have SVM reported in '/proc/cpuinfo' inside the guest VM is using 'host-passthrough' mode, but I want to avoid it to allow migration of the VM. Is this an issue with AMD This is part of the description of the guest: <domain type='kvm'> <name>vagrant-fedora-libvirt-nested_default</name> <uuid>4a379890-b891-4971-aad6-12fad45eaebc</uuid> <memory unit='KiB'>8388608</memory> <currentMemory unit='KiB'>8388608</currentMemory> <vcpu placement='static'>4</vcpu> <os> <type arch='x86_64' machine='pc-i440fx-2.3'>hvm</type> <boot dev='hd'/> </os> <features> <acpi/> <apic/> <pae/> </features> <cpu mode='custom' match='exact'> <model fallback='allow'>qemu64</model> <feature policy='require' name='svm'/> </cpu> Regards, Sladjan

1 0

[libvirt-users] vms not persistent after upgrading to f22?
by Rich Megginson 28 Aug '15

28 Aug '15

With f21 and earlier, I could just call virt-install to create VMs, and they would persist after reboot. After upgrading to f22, when I shut off a VM, it is removed, and has to be recreated. How can I get VMs to persist? Is there a virsh create or virt-install switch to do this?

3 3

[libvirt-users] Isolated networks && test lab
by Spanky Horawitz 28 Aug '15

28 Aug '15

I'm trying to migrate my VirtualBox test lab to KVM. In my VB configuration, I have several VM routers with 4 interfaces each. Each router has a bridged interface which connects them all together and to the outside world. I have servers/clients dispersed on each of the other private (Isolated) segments that use the routers as default gateways. In trying to make this work, I used Virtual Machine Manager to define 3x Virtual Networks (Isolated networks) and created a VM router which has 3 interfaces (192.168.x.1) on each Isolated network and a 4th interface using the br0 interface to get to the outside world. The router is able to connect to the outside world. Using Virtual Machine Manager, I also created "client" VM and connected it's single interface to one of the Isolated networks. While testing connectivity of the "client" VM to the router, it seems to go straight to the Host (tested using 'ssh 192.168.x.1'.) I'm trying to isolate the networks from the Host and have the router receive packets for the 192.168.x.1 address on the individual Isolated network segments. Host: br0 (em1): 192.168.1.16 vibr0 (default): 192.168.122.1 virbr1: 192.168.2.1 virbr2: 192.168.3.1 virbr3: 192.168.4.1 VM router: eth0: 192.168.1.10 eth1: 192.168.2.1 eth2: 192.168.3.1 eth3: 192.168.4.1 VM client: eth0: 192.168.3.101 Is there some trick to isolate the Host from the VM's? Is there a comprehensive guide that explains how the networks should be setup (with/without Virtual Machine Manager)? I'm not afraid to get my hands dirty on the console if needed, but GUI's are easier to navigate when you're learning a new technology. I switched to using RemoteBox for managing my VB vm's as it made things a lot easier to deal with than VBoxManage. Thanks for your help!

2 10

Re: [libvirt-users] Using unix-type channel ....
by Michal Privoznik 28 Aug '15

28 Aug '15

On 28.08.2015 08:34, Jean-Pierre Ribeauville wrote: > HI, > > > > Thx for your hint. > > > > I've tried this instead : > > > > <channel type='unix'> > > <source mode='bind' path='/opt/omnivision_1.agent'/> > > <target type='virtio' name='omnivision_1.agent'/> > > <address type='virtio-serial' controller='0' bus='0' port='2'/> > > </channel> > > > > Then, when starting the Guest I get this error message : [It's better to send just text via e-mail] So, the error message is telling you, that qemu process does not have enough permissions to bind to the unix socket under the given path. Either you should level up qemu permissions (e.g. run it as root), or use path that qemu will have access to. Or set correct permissions on the socket target. Michal

2 1

[libvirt-users] Using unix-type channel ....
by Jean-Pierre Ribeauville 28 Aug '15

28 Aug '15

Hi, I'm trying to use unix-type channel to send/receive datas between host and guest . Here are the steps I follow : 1) Adding these lines to expose a channel unix type to my guest <channel type='unix'> <source mode='bind' path='/var/lib/libvirt/qemu/omnivision_1.agent'/> <target type='virtio' name='org.qemu.guest_agent.0'/> <alias name='channel0'/> <address type='virtio-serial' controller='0' bus='0' port='2'/> </channel> 2) the VM_Orion-1 Guest, exposing a channel "unix" omnivision_1.agent is started, then netstat shows these new things : netstat -apn | grep qemu unix 2 [ ACC ] STREAM LISTENING 37214 19124/qemu-kvm /var/lib/libvirt/qemu/VM_Orion-1.monitor unix 2 [ ACC ] STREAM LISTENING 37219 19124/qemu-kvm /var/lib/libvirt/qemu/omnivision_1.agent unix 3 [ ] STREAM CONNECTED 37242 19124/qemu-kvm /var/lib/libvirt/qemu/omnivision_1.agent unix 3 [ ] STREAM CONNECTED 37240 19124/qemu-kvm /var/lib/libvirt/qemu/VM_Orion-1.monitor 3) I start , on the host , a simple test in order to send datas to the guest: ..... sock = socket(AF_UNIX, SOCK_STREAM, 0); if (sock < 0) { perror("opening stream socket"); return(1); } server.sun_family = AF_UNIX; strcpy(server.sun_path, argv[1]); printf("connecting to socket number : %d \n",sock); fflush(stdout); if (connect(sock, (struct sockaddr *) &server, sizeof(struct sockaddr_un)) < 0) { close(sock); perror("connecting stream socket"); return(1); } printf("writing data on stream socket"); if (write(sock, DATA, sizeof(DATA)) < 0) perror("writing on stream socket"); close(sock); ..... 4) invoking > test "/var/lib/libvirt/qemu/omnivision_1.agent" then I got : connecting to socket number : 3 writing data on stream socket and following netstat result : netstat -apn | grep qemu unix 2 [ ACC ] STREAM LISTENING 39348 20587/qemu-kvm /var/lib/libvirt/qemu/VM_Orion-1.monitor unix 2 [ ACC ] STREAM LISTENING 39353 20587/qemu-kvm /var/lib/libvirt/qemu/omnivision_1.agent unix 2 [ ] STREAM CONNECTING 0 - /var/lib/libvirt/qemu/omnivision_1.agent unix 3 [ ] STREAM CONNECTED 39375 20587/qemu-kvm /var/lib/libvirt/qemu/omnivision_1.agent unix 3 [ ] STREAM CONNECTED 39373 20587/qemu-kvm /var/lib/libvirt/qemu/VM_Orion-1.monitor 5) if I try to re-invoke test "/var/lib/libvirt/qemu/omnivision_1.agent" then : netstat -apn | grep qemu tcp 0 0 127.0.0.1:5902 0.0.0.0:* LISTEN 20587/qemu-kvm tcp 0 0 127.0.0.1:5902 127.0.0.1:57772 ESTABLISHED 20587/qemu-kvm unix 2 [ ACC ] STREAM LISTENING 39348 20587/qemu-kvm /var/lib/libvirt/qemu/VM_Orion-1.monitor unix 2 [ ACC ] STREAM LISTENING 39353 20587/qemu-kvm /var/lib/libvirt/qemu/omnivision_1.agent unix 2 [ ] STREAM CONNECTING 0 - /var/lib/libvirt/qemu/omnivision_1.agent unix 2 [ ] STREAM CONNECTING 0 - /var/lib/libvirt/qemu/omnivision_1.agent unix 3 [ ] STREAM CONNECTED 39375 20587/qemu-kvm /var/lib/libvirt/qemu/omnivision_1.agent unix 3 [ ] STREAM CONNECTED 39373 20587/qemu-kvm /var/lib/libvirt/qemu/VM_Orion-1.monitor on the guest side , I try to retrieve datas sent by host , by using this kind of thing : ... int fd = open("/dev/virtio-ports/org.qemu.guest_agent.0", O_RDWR | O_NONBLOCK); ssize_t size = read(fd, &buffer, 100); ... As you can imagine , I never receive any data from the host . Is it the right way to use this "unix-type" channel ? Thx for your patience. Regards, J.P. Ribeauville P: +33.(0).1.47.17.27.87 Puteaux 3 Etage 5 Bureau 4 jpribeauville(a)axway.com<mailto:jpribeauville@axway.com> http://www.axway.com<http://www.axway.com/> P Pensez à l'environnement avant d'imprimer.

2 2

[libvirt-users] Virtio serial exposition
by Jean-Pierre Ribeauville 27 Aug '15

27 Aug '15

Hi, I try to exchange datas between host and Guest by adding these lines in a guest .xml conf file : <channel type='pty'> <target type='virtio' name='arbitrary.virtio.serial.port.name'/> <address type='virtio-serial' controller='0' bus='0' port='1'/> </channel> Then /dev/virtio-ports/arbitrary.virtio.serial.port.name is exposed in the guest and a /dev/pts/x is seen as source path on the host. I checked that on two Linux hosts : 1) libvirt-daemon-driver-qemu-1.2.8-16.el7_1.3.x86_64 qemu-kvm-1.5.3-86.el7_1.5.x86_64 qemu-kvm-common-1.5.3-86.el7_1.5.x86_64 qemu-guest-agent-2.1.0-4.el7.x86_64 ipxe-roms-qemu-20130517-6.gitc4bce43.el7.noarch qemu-img-1.5.3-86.el7_1.5.x86_64 Linux sparring 3.10.0-229.el7.x86_64 #1 SMP Thu Jan 29 18:37:38 EST 2015 x86_64 x86_64 x86_64 GNU/Linux 2) qemu-img-0.12.1.2-2.415.el6.x86_64 qemu-kvm-0.12.1.2-2.415.el6.x86_64 gpxe-roms-qemu-0.9.7-6.10.el6.noarch Linux orion 2.6.32-431.el6.x86_64 #1 SMP Sun Nov 10 22:19:54 EST 2013 x86_64 x86_64 x86_64 GNU/Linux The device is only exposed on the guest of the first host . Is it due to a too old version of qemu and/or Linux on the second host or some packages missing on this second host ? Thx. Regards, J.P. Ribeauville P: +33.(0).1.47.17.27.87 Puteaux 3 Etage 5 Bureau 4 jpribeauville(a)axway.com<mailto:jpribeauville@axway.com> http://www.axway.com<http://www.axway.com/> P Pensez à l'environnement avant d'imprimer.

2 2