- Users - libvirt List Archives

[libvirt-users] Quick questions on virtio-serial transport
by Hari Pyla 01 Aug '13

01 Aug '13

Hi, I am using the virtio-serial transport for communication between the host and the guest. I've created the channel using virt-install: virt-install --name=centos --import --disk /var/lib/libvirt/images/centos.img,size=15 \ --ram 2048 \ --vcpus=2 \ --arch=x86_64 --os-type linux --network=bridge:br1 --graphics vnc \ --noautoconsole \ --channel pty,target_type=virtio,name=guest1 This correctly creates a file "/dev/virtio-ports/guest1" in the guest for the guest to communicate with the host. So the guest side of things are good. Now on the host side, I see a two files being created in /dev/pts directory. $ ls -alt /dev/pts crw--w---- 1 qemu tty 136, 1 Jul 31 15:49 1 crw--w---- 1 qemu tty 136, 2 Jul 31 15:49 2 So here are my questions: i) I was wondering if there is anyway of knowing which of these files correspond to the other end of the channel without having to manually echoing data to each of these files and see which of the data shows up on the guest side. ii) I was also wondering if there is anyway of identifying the host side endpoint i.e., /dev/pts/FILE of the channel programatically iii) Better yet can the /dev/pts/FILE be explicitly passed as an argument while creating a channel in the guest so the host knows that it is supposed to use the /dev/pts/FILE if it wants to talk to the guest. I read a prior post on this mailing list: https://www.redhat.com/archives/libvir-list/2012-May/msg00964.html It mentions that from the host side one could dump the xml of the guest and identify the "source path" of the channel, something in the lines of <channel type='pty'> <source path='/dev/pts/12'/> <target type='virtio' name='arbitrary.virtio.serial.port.name'/> <alias name='channel0'/> <address type='virtio-serial' controller='0' bus='0' port='2'/> </channel> but I do not find the source path in the XML description of the guest I've created. This is what I see in the XML file. <channel type='pty'> <target type='virtio' name='guest1'/> <address type='virtio-serial' controller='0' bus='0' port='1'/> </channel> I was wondering if I am missing something. Any help is greatly appreciated. Thanks, --Hari

1 0

[libvirt-users] filesystem accessmode='passthrough'
by Yury Goltsov 31 Jul '13

31 Jul '13

Hello. I am trying to 'passthrough' the part of host filesystem to kvm guest. libvirt xml: <filesystem type='mount' accessmode='passthrough'> <source dir='/var/guests/mail-var'/> <target dir='mail-var'/> </filesystem> kvm is started from root. After mounting in guest mount -t 9p mail-var /mnt -o trans=virtio,version=9p2000.L or mount -t 9p www-var /mnt -o trans=virtio if I try to make file or directory in guest, I have problem with permissions: ruser@www:~# ls -ld /mnt/ruser/ drwxr-xr-x 2 ruser www-data 4096 May 28 13:13 /mnt/ruser/ ruser@www:~# ls -ld /mnt/tmp/ drwxrwxrwx 2 root root 4096 Jul 26 09:01 /mnt/tmp/ root@www:~# touch /mnt/tmp/file1 OK: root@www:~# touch /mnt/ruser/file2 touch: cannot touch `/mnt/ruser/file2': Permission denied root@www:~# su ruser ruser@www:~$ touch /mnt/ruser/file3 touch: cannot touch `/mnt/ruser/file3': Permission denied ruser@www:~$ touch /mnt/tmp/file4 touch: setting times of `/mnt/tmp/file4': No such file or directory I have read in documentation, that in "passthrough" security model, files are stored using the same credentials as they are created on the guest (like NFS3). Suggest me, please, what is it: I do something wrong, or I must update qemu-kvm or/and libvirt to recent version, or I need to find other solution for sharing host filesystem? Thanks. Yury in host: root@host:~# libvirtd --version libvirtd (libvirt) 0.9.12 root@host:~# kvm --version QEMU emulator version 1.1.2 (qemu-kvm-1.1.2+dfsg-6, Debian), Copyright (c) 2003-2008 Fabrice Bellard root@host:~# uname -a Linux host 3.2.0-4-amd64 #1 SMP Debian 3.2.46-1 x86_64 GNU/Linux root@host:~$ cat /etc/os-release PRETTY_NAME="Debian GNU/Linux 7 (wheezy)"

2 5

[libvirt-users] Fail to destroy lxc container: internal error Some container PIDs refused to die
by hzguanqiang 31 Jul '13

31 Jul '13

Hey guys, When I destroy a lxc container, It reported an error with details as following: root@debian:~$ vir destroy 901eef81-06b2-4252-aeef-53b87c402585 error: Failed to destroy domain 901eef81-06b2-4252-aeef-53b87c402585 error: internal error Some container PIDs refused to die root@debian:~$ vir version Compiled against library: libvirt 1.1.0 Using library: libvirt 1.1.0 Using API: LXC 1.1.0 Running hypervisor: LXC 3.2.46 What's the problem? How could I destroy the lxc just like this? Thanks. -------------- Best regards! GuanQiang 2013-07-31

2 1

Re: [libvirt-users] Installing libvirt
by Peeyush Gupta 31 Jul '13

31 Jul '13

Hi, I am not sure what exactly "libvirt-related packages" mean. But I didn't have anything else on my system. I mean, I launch a VM (with Ubuntu 12.04) and the first thing I installed is libvirt only. Thanks. ~Peeyush Gupta

1 0

[libvirt-users] lxc-enter-namespace error: security model cannot be entered.
by hzguanqiang 30 Jul '13

30 Jul '13

Hi Guys, I started a lxc container with libvit in ubuntu Operating system, and succeed using lxc-enter-namespace to enter the namespaces and security context of the container. But when I do the same thing in debian OS, It reported an error, with details as following: root@debian:/etc# vir list Id Name State ---------------------------------------------------- 4424 instance-00000007 running 25913 instance-00000008 running root@debian:/etc# vir dumpxml 4424 <domain type='lxc' id='4424'> <name>instance-00000007</name> <uuid>f1ce5360-bb5e-4cfc-b5ef-d05f8db52618</uuid> <memory unit='KiB'>1048576</memory> <currentMemory unit='KiB'>1048576</currentMemory> <vcpu placement='static'>3</vcpu> <resource> <partition>/machine</partition> </resource> <os> <type arch='x86_64'>exe</type> <init>/sbin/init</init> <cmdline>console=tty0 console=ttyS0</cmdline> </os> <clock offset='utc'/> <on_poweroff>destroy</on_poweroff> <on_reboot>restart</on_reboot> <on_crash>destroy</on_crash> <devices> <emulator>/usr/lib/libvirt/libvirt_lxc</emulator> <filesystem type='mount' accessmode='passthrough'> <source dir='/opt/stack/data/nova/instances/f1ce5360-bb5e-4cfc-b5ef-d05f8db52618/rootfs'/> <target dir='/'/> </filesystem> <interface type='bridge'> <mac address='fa:16:3e:3a:c6:11'/> <source bridge='br100'/> <target dev='veth0'/> <filterref filter='nova-instance-instance-00000007-fa163e3ac611'/> </interface> <console type='pty' tty='/dev/pts/1'> <source path='/dev/pts/1'/> <target type='lxc' port='0'/> <alias name='console0'/> </console> </devices> <seclabel type='none'/> </domain> root@debian:/etc# vir lxc-enter-namespace 4424 /bin/sh/ libvirt: error : argument unsupported: Security model cannot be entered Is there anything that needs to be configured in debian OS for using the 'lxc-enter-namespace' interface? -------------- Best regards! GuanQiang 2013-07-30

2 2

[libvirt-users] ANNOUNCE: Oz 0.11.0 release
by Chris Lalancette 30 Jul '13

30 Jul '13

All, I'm pleased to announce release 0.11.0 of Oz. Oz is a program for doing automated installation of guest operating systems with limited input from the user. Release 0.11.0 is a bugfix and feature release for Oz. Some of the highlights between Oz 0.10.0 and 0.11.0 are: * Add support for installing Ubuntu 13.04 * Add the ability to get user-specific ICICLE information * Add the ability to generate ICICLE safely, by using a disk snapshot * Add the ability to include extra files and directories on the installation ISO * Add the ability to install to alternate file types, like qcow2, etc. * Add support for installing Ubuntu 5.04/5.10 * Add support for installing Fedora 19 * Add support for installing Debian 7 * Add support for Windows 2012 and 8 * Add support for getting files over http for the commands/files section of the TDL * Add support for setting a custom MAC address to guests during installation * Add support for user specified disk and NIC model * Add support for OpenSUSE 12.3 * Add support for URL based installs for Ubuntu A tarball and zipfile of this release is available on the Github releases page: https://github.com/clalancette/oz/releases . Packages for Fedora-18 and Fedora-19 have been built in Koji and will eventually make their way to stable. Instructions on how to get and use Oz are available at http://github.com/clalancette/oz/wiki . If you have questions or comments about Oz, please feel free to contact me at clalancette at gmail.com, or open up an issue on the github page: http://github.com/clalancette/oz/issues . This was one of the most active Oz releases ever, because of the feedback and patches from the community. Thanks to everyone who contributed to this release through bug reports, patches, and suggestions for improvement. Chris Lalancette

2 2

[libvirt-users] Libvirt and Glusterfs pool
by Pierre-Gilles Mialon 28 Jul '13

28 Jul '13

Hi, I use the QEMU-GlusterFS native integration (no Fuse mount) with the libvirt. Now I create a volume issuing : # qemu-img create gluster://localhost/gv1/test.img 5G Then using the libvirt I declare the following lines in my domain.xml : <disk type='network' device='disk'> <driver name='qemu' cache='none'/> <source protocol='gluster' name='gv1/test.img'> <host name='127.0.0.1' transport='tcp' /> </source> <target dev='vda' bus='virtio'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x05' function='0x0'/> </disk> It works really well. But the better way, would be to only use the libvirt. Is it planned to support a native glusterFS pool type ? That would be great to be able to define a pool like this: <pool type='glusterfs'> <name>myname</name> <source protocol='gluster' volume='gv0'> <host name='127.0.0.1'/> </source> <target> <permissions> <mode>0700</mode> <owner>-1</owner> <group>-1</group> </permissions> </target> </pool> And then create new images using vol-create. Regards, -- Pierre-Gilles Mialon

2 1

[libvirt-users] pinVcpu not working
by Peeyush Gupta 27 Jul '13

27 Jul '13

Hi all, I am working with libvirt and I am trying to set cpu affinity. Now I can always use virsh vcpupin <domain_name> <vcpu> <pcpu> to pin vcpus to pcpus. I want to do it using Python API. Now, there is function pinVcpu which is supposed to do that. But this is not working. For example I gave dom.pinVcpu(0,1) but still my vcpu affinity is for all the pcpus. The function returns 0 (success). Any idea what am I doing wrong? Thanks. ~Peeyush Gupta

2 2

[libvirt-users] dual head with virt-manager, kvm, debian
by Thomas Koch 26 Jul '13

26 Jul '13

Hi, I try to run a debian testing guest inside a debian stable host and would like to use both physical displays to show virtual displays of my guest. Is this possible? I tried setting the heads attribute in the domain xml as described here: https://bugzilla.redhat.com/show_bug.cgi?id=885806#c1 I've tried virt-managers internal viewer, virt-viewer and spicy without any indication of a second display. Regards, Thomas Koch

1 0

[libvirt-users] How to trigger a script in a guest after resume? (aka guest system clock incorrect after host reboot with suspended guest)
by Nils Toedtmann 24 Jul '13

24 Jul '13

Hi Is there a way i can trigger a script within a libvirt guest immediately after resume? E.g. i don't see any log message that indicates to a guest that it had been paused and was just woken up. The problem that i am trying to solve is that when i reboot the host (pausing all guests), the guests' system clocks are a few minutes late after resume. The guests RTC (current_clocksource = kvm-clock) are correct though, so a "hwclock --hctosys" fixed the offset. I only need to find out how to call hwclock right after resume! Right now i call hwclock every few minutes, but i'd like to shorten the window of wrong time. OTOH i'd like to avoid installing ntpd into all guests, given their perfectly fine RTCs. On the host i am using CentOS 6.4 with all patches and the packaged libvirt-0.10.2 and qemu-kvm-0.12.1.2. All guests are Ubuntu 12.04. /nils.

2 1