- Users - libvirt List Archives

[libvirt-users] Allocating Sockets Cores and Threads.
by joe fu 27 Jan '11

27 Jan '11

Hi, Is there any other way I could allocate sockets, cores and threads to my virtual machine? I looked over the virsh man page and I didn't find anything that talks about sockets, threads and cores. I edited the vm xml file and added this... <cpu> <topology sockets='1' cores='2' threads='1'/> </cpu> Saw this in the libvirt documents, http://libvirt.org/formatdomain.html Setting this up shows me a single core in my vm, but I don't see any sockets and threads. Can somebody tell me how to do it in a right way so that I see all sockets, cores and threads allocated in my vm. Thanks joe

2 3

[libvirt-users] Is there a way to suppress netfilter rules (default.xml)?
by Oliver Schneider 27 Jan '11

27 Jan '11

Hi, in order to get hook scripts to work, I used a backported version of libvirtd on Ubuntu 10.04 LTS. Here are the details: Compiled against library: libvir 0.8.3 Using library: libvir 0.8.3 Using API: QEMU 0.8.3 Running hypervisor: QEMU 0.12.5 After adjusting the apparmor profile to accept the hook scripts under /etc/libvirt/hooks/, I am able to start and stop domains again, but only the script named "daemon" gets executed, the one named "qemu" gets ignored. All my domains are KVM domains. I thought the "qemu" script should apply to that?! My actual goal is to suppress the rules or better yet execute my own script (in order to amend or replace the libvirt rules) after insertion of the "default" rules during startup of the daemon. The reason being that I need to do some PNAT and other custom handling ... So far I have not been able to catch the state where the bridge has already been created and the rules inserted. Thus I've had to manipulate the firewall rules manually (well, still scripted ;)) after booting the host. Any ideas? Thanks, // Oliver PS: Side-note, I tried using the newfilter facilities, but there seems to be no way to insert rules at the start of a chain and similar things. PPS: # cat /etc/libvirt/qemu/networks/default.xml <network> <name>default</name> <bridge name="virbr%d" /> <forward/> <ip address="192.168.122.1" netmask="255.255.255.0" /> </network>

1 0

[libvirt-users] invalid argument error when using cache parameter
by Igor Serebryany 26 Jan '11

26 Jan '11

hi, i tried to add a 'cache' argument to the 'driver' element in my domainxml for qemu/kvm. i tried adding cache="none" as well as "writethrough" and "writeback", but every time i get these errors: libvirtError: internal error process exited while connecting to monitor: char device redirected to /dev/pts/1 qemu: could not open disk image /vms/test.img: Invalid argument the full driver line looks like this: <driver name="qemu" type="raw" cache="none"/> the same domainxml works just fine without the cache parameter. does anyone know what i'm doing wrong? is there some other way to set the default cache policy on qemu/kvm? i didn't see anything in the config file... --igor

3 4

[libvirt-users] git repository access via HTTPS
by Fyodor Kupchik 24 Jan '11

24 Jan '11

Hi, I've just tried to do a fresh checkout (git clone) with libvirt sources tree and had no luck because it seems to be inaccessible/non-existent at all. Is this temporary situation or I have to only use git:// proto? I'm asking because I'm sitting behind rather restrictive corporate proxy which does not allow anything but a little set of protocols such as HTTP(S). Thanks.

6 17

[libvirt-users] libvirt and shared storage SAN Fiber Channel
by Frederic SOULIER 24 Jan '11

24 Jan '11

Hi, I'm looking information about using libvirt with san (fiber channel) based storage and googling for a while i don't see anything about it. I send this email in order to get advices about libvirt and shared storage We use here, for 3 years now, 8 linux centos server connected to an hitachi FC SAN (multipath devices). Each server run Xen dom0 and use SAN LUNs to store Virtual machine filesystem ( basically one LUN per VM). For each LUN we create an LVM VolGroup. On each VolGroup we create Logical volumes (one for swap, one for root filesystem). We present to the vm the 2 Logical Volume as it was raw disk. In this system we can make live migration between host using local scripts which assume than one VolumeGroup is used only by one host at the same time on the 8 node "cluster". We don't use any clustered aware environment like CLVM. We plan to migrate to kvm and libvirt with the new centos release (CentOS 6). Reading the upstream documentation and libvirt i don't find anything about using libvirt with san storage except for iscsi. Do you have any link or idea about using kvm + libvirt + SAN FC storage that permit to make live migration between node (we think about RHCS with GFS2 but it seems very complex to set up) ? Do you think i can use the LUN's SAN as local storage on each node. For each LUN create an LVM Volume Group (same on each node, like that we doing now with xen) and use it as volume pool in libvirt. Is volume pool in libvirt is 'cluster aware' (Can i access it simultaneously from different node) ? I'm beginner in kvm so any advice will be great. regards -- Frederic Soulier DSI / Service Système Université Toulouse 1 Capitole 2 rue du doyen Gabriel Marty 31 042 Toulouse Cedex 9 Tel: +33 5 61 63 39 98 Fax: +33 5 61 63 37 98

1 1

[libvirt-users] Cgroup
by Nikita A Menkovich 24 Jan '11

24 Jan '11

Hello, I'm running Debian Sid with libvirt/qemu/kvm packages from experimental: $ qemu --version QEMU emulator version 0.13.0 (Debian 0.13.0+dfsg-2), Copyright (c) 2003-2008 Fabrice Bellard $ kvm --version QEMU emulator version 0.13.0 (qemu-kvm-0.13.0 Debian 0.13.0+dfsg-2), Copyright (c) 2003-2008 Fabrice Bellard $ libvirtd --version libvirtd (libvirt) 0.8.6 /etc/libvirt/qemu.conf cgroup_controllers = [ "cpu", "devices" ] cgroup_device_acl = [ "/dev/null", "/dev/full", "/dev/zero", "/dev/random", "/dev/urandom", "/dev/ptmx", "/dev/kvm", "/dev/kqemu", "/dev/rtc", "/dev/hpet", "/dev/net/tun", ] /etc/cgconfig.conf mount { cpu = /dev/cgroup/cpu; cpuacct = /dev/cgroup/cpuacct; devices = /dev/cgroup/devices; # memory = /dev/cgroup/memory; blkio = /dev/cgroup/blkio; } When I'm running virsh, and want to change cpu shares I receive an error virsh # schedinfo --set cpu.shares=2048 test Scheduler : posix error: internal error cannot find cgroup for domain test Really I didn't have any group test in $ ls /dev/cgroup/cpu/sysdefault/libvirt/qemu/ cgroup.clone_children cgroup.event_control cgroup.procs cpu.shares notify_on_release tasks It happens if I run libvirt as usual user, how can I grant access to create cgroup to ordinary user? -- Nikita A Menkovich JID: menkovich(a)gmail.com

2 1

[libvirt-users] Python Bindings for Python 2.7
by vipul borikar 20 Jan '11

20 Jan '11

Hello, I want to source compile the libvirt 0.8.4 with its Python bindings. I have python 2.4 rpm installed but i dont use it. I use Python 2.7 source compiled version. So when i compile libvirt it produces bindings for Python 2.4 which is obvious choice since it is rpm. So how should i make it point to Python 2.7 so that it produces bindings for it. Installed path of Python 2.7 is /opt/Python2.7. Please Guide me. Thanks Vipul

3 3

[libvirt-users] virt-install with --channel option inquiry
by Hongqing Yang 20 Jan '11

20 Jan '11

hi, virt guys, This is Hongqing from Fedroa QA team. I try to forward the guest installation logs to host with virtio. I have used virsh edit <guestName> to add a channel, it works fine. I think it would be better if I can initialize it when I create the guest using virt-install, and virt-install also offers the option, I have tried below, but it does not work. virt-install --connect qemu:///system \ --virt-type kvm \ --name demo2 \ --ram 500 \ --disk path=/home/hongqing/testscript/demo2.img,size=10,sparse=true \ --accelerate \ --cdrom /home/hongqing/fedora/Fedora-14-x86_64-DVD.iso \ --channel tcp host=127.0.0.1:6080 mode=connect target_type=virtio name=org.fedoraproject.anaconda.log.0 error message is : ERROR internal error Process exited while reading console log output: char device redirected to /dev/pts/20 getaddrinfo(127.0.0.1,None): Servname not supported for ai_socktype chardev: opening backend "socket" failed both the host and guest are fedaora 14. I have checked the service syslog with port 6080 is listening. and also I have disabled the SElinux. Do you have any idea about this? Thanks for taking time to view it. Hongqing YANG

2 4

[libvirt-users] Getting CPU and memory usage statistics for the host
by Guido Winkelmann 19 Jan '11

19 Jan '11

Hi, I have noticed that virt-manager has the ability to show me the current cpu- and memory-usage of both the host and running guest machines. How does it do that? I could make good use of that information myself in the client applications I'm writing, however, from what I can see on http://libvirt.org/html/libvirt-libvirt.html, I have no idea how to go about getting it from libvirt. In fact, before I've seen virt-manager, I just assumed libvirt cannot do that at all, and was planning to manually ssh to the host and read statistics from /proc when I'd need them. I've tried looking at the virt-manager source code to find out how to do this, but it appears I would have to learn Python first in order to understand it. Could someone please give me some pointers? Regards, Guido

2 1

[libvirt-users] host to VM serial device configuration
by John Paul Walters 19 Jan '11

19 Jan '11

Hi, I'm trying to get a virtual serial device up and running between my host and virtual machine with a device name on the host. I'm using libvirt 0.8.3 and qemu 0.13.0. The challenge that I'm running into is that I'm able to get a serial device, but I cannot fix it to a pre- defined device name. For example, I'm using the following in my VM's xml file: <serial type='pty'> <source path='/dev/pts/19' /> <target port='0' /> </serial> As I said this works, but it doesn't set the host side to /dev/pts/ 19. Is there any way to do this? thanks, JP

5 7