- Users - libvirt List Archives

[libvirt-users] Libvirt heavy CPU usage
by Samuel Hassine 04 May '12

04 May '12

Hi there, We are currently using OpenStack with the libvirt driver (libvirt version: 0.9.11.3) and LXC containers. We are observing a really bad CPU usage of libvirt_lxc processes: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 28828 root 20 0 40852 1376 832 R 104.4 0.0 516:44.27 libvirt_lxc 20560 root 20 0 40852 1372 832 R 98.6 0.0 454:42.48 libvirt_lxc 22419 root 20 0 40852 1376 832 R 98.6 0.0 499:16.98 libvirt_lxc 26238 root 20 0 40852 1376 832 R 98.6 0.0 500:42.94 libvirt_lxc 28307 root 20 0 40852 1376 832 R 92.8 0.0 594:16.96 libvirt_lxc 20408 root 20 0 40852 1376 832 R 69.6 0.0 364:21.02 libvirt_lxc 17495 root 20 0 40852 1372 832 R 63.8 0.0 421:40.78 libvirt_lxc 23725 root 20 0 40852 1376 832 R 46.4 0.0 459:03.75 libvirt_lxc 24710 root 20 0 40852 1368 832 R 40.6 0.0 494:30.45 libvirt_lxc 1350 root 20 0 40852 1372 828 R 34.8 0.0 358:49.72 libvirt_lxc But theses domains are not really active, they are just "minimal" virtual engines without any trafic or load. Is it a "normal" way or there is something wrong? Thanks for your answer. Best regards. Sam

1 0

[libvirt-users] free memory on hypervisor
by John Wayne 04 May '12

04 May '12

i am trying to get the free memory on a hypervisor using libvert, with the python api bindings as per http://libvirt.org/python.html, it should be pretty easy, however the below is failing: #conn is a connection object freemem=libvirt.virNode.getFreeMemory(conn) although this works fine: software=libvirt.virConnect.getType(conn) if software == 'QEMU': logging.debug('hypervisor is type QEMU')

2 1

[libvirt-users] communicate with vm
by me,apporc 03 May '12

03 May '12

We need to create many virtual machines and assign an static ip for each of them. To do this, 1. we can use one dhcp server to assign ip for them, and bind the MAC to it's ip. But we got many many machines, so this is not easy to operate for admins. 2. we can just create the vm and then let the client to config the ip from within the operating system. But the client don't want to, and sometimes in fact they should not be authorized to do that. Are there any way else to do this? I mean is there some API for that we can use? or just mount the vm image file and change the network configuration file? If any API for that, can we change the config while the os in the vm is running? Thanks, any help is appreciated.

2 1

[libvirt-users] ESX support for virt-install
by Filip Skola 03 May '12

03 May '12

Hi! I try to integrate ESXi guests to beaker. I'd like to have dynamic installations of guests just like KVM and Xen is done. I know that virt-install is not currently supported by libvirt, so I'd like to ask, do you plan to add esxdrv support for virt-install? What's current status? Thank you, Filip

1 0

[libvirt-users] USB passthrough: flexible configuration in XML-file needed
by eGerlach 03 May '12

03 May '12

Hi, I need a flexible USB passthrough of a USB-device. If the device is turned off by the current switch (230Volt) the VM should start and not claim that the USB device is left. Using: <devices> … <hostdev mode='subsystem' type='usb'> <source> <vendor id='0x152a'/> <product id='0x8180'/> </source> </hostdev> </devices> </domain> in libvirt XML-config file the boot process is stopped if the device is switched off. How can I configure the xml-file to make ich more flexible? Another possibility would be to passthrough USB-Slot No1 (or No 2, ...) to the VM. So an arbitrary USB-device that is plugged in to USB-Slot1 would be recognized by the VM. In another forum I got the hint to start the VM by: -usb -usbdevice host:0.* -usbdevice host:1.* to passthrough USB-Slot No1 and No 2. How can I configure that in libvirt xml-file? tia Eckard

1 0

[libvirt-users] Regarding persistence of VM's after live migration (virDomainMigrateToURI() problem)
by Coding Geek 02 May '12

02 May '12

Hello I am working with 3 host machines each running xen with shared NFS storage. I am working on automatic load balancing if one host is over utilized and another is under utilized by measuring the utilization from xentop. I am facing a problem after migration of VM. I am setting the flags ( 1| 8| 16) in order to do live migration, persist VM on destination, undefine host from source. After migration if I shut off the migrated VM on destination host it does not persist. I am using the migrateToURI() API for migration. Please help how to make VM persist on destination. Thank you, Tasvinder Singh -- -- "To Iterate is Human, To Recurse is Divine"

3 6

[libvirt-users] Snooze energy-efficient cloud manager is available as open-source!
by Eugen Feller 02 May '12

02 May '12

Dear all, herewith we proudly announce the first public release of Snooze. Snooze is an open-source scalable, autonomic, and energy-efficient virtual machine (VM) management framework for private clouds based on libvirt. Similarly to other VM management frameworks such as Nimbus, OpenNebula, Eucalyptus, and OpenStack it allows to build compute infrastructures from virtualized resources. Particularly, once installed and configured users can submit and control the life-cycle of a large number of VMs. However, contrary to existing frameworks for scalability and fault tolerance, Snooze employs a self-organizing and healing hierarchical architecture. Moreover, it performs distributed VM management and is designed to be energy efficient. Therefore, it implements features to monitor and estimate VM resource (CPU, memory, network Rx, network Tx) demands, detect and resolve overload/underload situations, perform dynamic VM consolidation through live migration, and finally power management to save energy. Last but not least, it integrates a generic scheduler which allows to implement any VM placement algorithms. The system can be either used to manage production data centers or as an experimental testbed for advanced (i.e. requiring live migration support) VM placement algorithms. Please see our website "http://snooze.inria.fr" for more information. We are looking forward for your feedback and contributions! Best regards, Eugen -- Eugen Feller PhD student, University of Rennes I INRIA Rennes Bretagne Atlantique - MYRIADS project Tel: +33 (0) 2 99 84 72 68 Web: http://www.irisa.fr/myriads/members/efeller

2 1

[libvirt-users] Python Libvirt Bindings
by John Wayne 02 May '12

02 May '12

Hi, Can anyone please share sample Python code for using qemu+tcp with SASL or qemu+tls with SASL? (i.e., connecting to a hypervisor using qemu+tls or qemu+tcp when a username/password using SASL has been defined on the hypervisor) I am having some trouble figuring out how to use callbacks for authenticating from with my Python program. I assume I will have to use http://libvirt.org/html/libvirt-libvirt.html#virConnectCredentialType, but I can't quite translate that to Python from C. Thanks

2 1

[libvirt-users] serial console locking for a domain (concurrent access = garbled output)
by Dusty Mabe 01 May '12

01 May '12

Hi, I am seeing an issue in libvirt-0.9.4-23.el6_2.7.x86_64.rpm where my virsh console output for a domain gets garbled because I have virt-manager running and it is also trying to access the serial device for the guest domain. Basically there is concurrent serial access and half the text ends up on one end while the rest ends up on the other end. Closing virt-manager does not help. Disconnecting and reconnecting using virsh console does not help. The only way to recover is to destroy the domain and then start the guest again and never start virt-manager. Note that the domain does not have VNC graphics so virt-manager defaults to try to access the serial port when started for a particular domain. Should there be locking within libvirt to only allow for a single client to access a specific serial port of a domain? Thanks, Dusty Mabe

2 2

[libvirt-users] Console to RHEL6.1 container
by Sukadev Bhattiprolu 01 May '12

01 May '12

We are trying to create RHEL6.1 container and having trouble getting a proper console to the container. I have used lxc containers in the past and lxc has a script (lxc-fedora) to setup a basic Fedora container. Is there something similar or docs with virtmgr ? We started out making a copy of the root fs and using that copy as the root for the container. I made a couple of tweaks to the container rootfs: 1. ensured that /dev/ptmx is a symlink to pts/ptmx 2. Added 'newinstance' option to the devpts mount in fstab The container xml file is attached. If devpts entry in container's fstab has the 'newinstance' option, I get a brief "domain test01 started" message on the stdout. When I run 'virsh consle test01', I only see a few messages --- Setting hostname node03: [ OK ] Setting up Logical Volume Management: No volume groups found [ OK ] Checking filesystems [ OK ] mount: can't find / in /etc/fstab or /etc/mtab Mounting local filesystems: [ OK ] Enabling local filesystem quotas: [ OK ] Enabling /etc/fstab swaps: [ OK ] --- and then a hang wihtout a shell. At ths point sshd or any of the other services don't appear to be running. If I remove the 'newinstance' entry in fstab and run 'virsh start test01' the container starts up ok. I get all "console" messages on whichever is my current /dev/pts/0 ! At this point sshd and many other services are running and I can ssh in. Even in this case, though 'virsh console test01' shows the same _few_ messages as above and no login shell. Back to the newinstance case, by putting some debug messages in the container's /etc/rc.sysinit, I see that devpts is already mounted with newinstance. With lxc, in setup_pts(), we unmount devpts in the container and then mount with newinstance. I wonder what that sequence is with virtmgr. I double checked that /dev/ptmx in container is a symlink to pts/ptmx and that CONFIG_DEVPTS_MULTIPLE_INSTANCES=y. Appreciate any tips on how I can get access to the console of the RHEL6.1 container. Suka Suka

1 1