June 2012 - Users - libvirt List Archives

[libvirt-users] Reconnection problem with a USB device
by Dale Amon 13 Jun '12

13 Jun '12

This is going to be a bit off the wall as I doubt many people are using a VM in quite the same manner as I am in this case. I have a new scanner at home, a Mustek A3USB 1200 Pro, which requires a vendor supplied version of libsane to run. That is a pain in the behind just to start with. To use that scanner requires a 32b machine with sufficiently old versions of various libraries. My primary machine, a laptop, is 64b and runs current software. So, in order to use my inexpensive large format scanner, I had to set up a 32b VM on my laptop that is running an old linux. I tunneled the USB device to it: <hostdev mode='subsystem' type='usb' managed='no'> <source> <vendor id='0x055f'/> <product id='0x040b'/> <address bus='3' device='8'/> </source> <alias name='hostdev0'/> </hostdev> That works fine... until I have to go into the office. Once I disconnected the USB go away for 10 hours and come back, it fails to re-engage. The USB seems to get grabbed up by the laptop itself instead of the VM. The only way to get it to work again is to halt the VM and do a cold boot of it. This is annoying to say the least because I lose 'state' when I boot, ie the set of windows and files open that remind me of where I left off on a project before going to work. Any suggestions how to get around this annoying issue?

2 4

[libvirt-users] What's the most recent compatible libvirt and qemu-kvm?
by Whit Blauvelt 13 Jun '12

13 Jun '12

Figured I'd test qemu-kvm-1.0.1 and libvirt-0.9.12, but the result after default compiles is: root@black:/usr/local/sbin# ./libvirtd 2012-06-12 19:53:41.894+0000: 6076: info : libvirt version: 0.9.12 2012-06-12 19:53:41.894+0000: 6076: warning : ebiptablesDriverInit:4084 : Could not find 'ebtables' executable 2012-06-12 19:53:42.675+0000: 6076: error : qemuCapsComputeCmdFlags:1209 : unsupported configuration: this qemu binary requires libvirt to be compiled with yajl 2012-06-12 19:53:42.677+0000: 6076: error : virDomainDefParseXML:8312 : internal error No guest options available for arch 'x86_64' Is it just the "yajl" thing missing, or is this combo just not to be expected to work? Also, should http://libvirt.org/compiling.html mention "yajl"? Thanks, Whit

4 5

[libvirt-users] How can a bridge be optimized?
by Whit Blauvelt 13 Jun '12

13 Jun '12

Hi, Netperf tells me that a KVM VM using a bridged connection to a system across the Gigabit LAN sees on average about 1/10th of the bandwidth the host sees. This is on very lightly loaded hosts, although with multiple low-use VMs. I've tried adjusting the CPU throttling after seeing that mentioned as a possible factor in slowing down a bridge, but it has not much if any effect here. Is there a good guide somewhere to what might make a difference with this? I recall the kernel recently adding a new alternative for handling VM interfaces, but can't recall what it was called. Would that be a marked improvement? Is it supported by libvirt? Thanks, Whit

6 17

[libvirt-users] Libvirt hangs until service is stopped and started again
by Ananth 13 Jun '12

13 Jun '12

Hi, The connection to libvirt seems to be hanging until the service is stopped and started again. But when i checked the status of libvirtd service (on Centos) and libvirt-bin service (on Ubuntu), service is still running. But any new connections and existing connections remain hung.I am using 0.9.4 version of libvirt on both the machines. I also found the following links reporting the same issue. https://bugzilla.redhat.com/show_bug.cgi?id=796451 http://www.redhat.com/archives/libvir-list/2009-June/msg00434.html Is this a known problem or is there a solution for this issue? Kindly let me know if there is a work around to avoid libvirt from hanging. Thank you. -- Regards Ananth

1 0

Re: [libvirt-users] How can a bridge be optimized?
by Whit Blauvelt 12 Jun '12

12 Jun '12

On Tue, Jun 12, 2012 at 11:35:28AM +0400, Andrey Korolyov wrote: > Just a stupid question: did you pin guest vcpus or NIC` hw queues? > Unpinned vm may seriously affect I/O performance when running on same > core set as NIC(hwraid/fc/etc). Thanks. No question is stupid. Obviously, this shouldn't be so slow at VM I/O so I'm missing something. In my defense, there is no single coherent set of documents on this stuff, unless those are kept in a secret place. It would be a fine thing if a few of the people who know all the "obvious" stuff about libvirt-based KVM configuration would collaborate and document it fully somewhere. When I Google "pinned vcpu" all the top responses are about Xen. I run KVM. I find mention that "KVM uses the linux scheduler for distributing workload rather than actually assigning physical CPUs to VMs," at http://serverfault.com/questions/235143/can-i-provision-half-a-core-as-a-vi…. Is that wrong? Are you suggesting RAID will end up on one CPU core, and if the VM ends up on the same core there's I/O trouble? I can see that for software RAID. But we're running hardware RAID. Isn't that handled in the hardware of the RAID controller? Isn't that the point of hardware RAID, to keep it off the CPU? In any case the slow IO is true of all the VMs, so presumably they're properly spread out over CPU cores (of which there are more than VMs, and each VM is configured to take only one), which would rule out the general problem being the shared use of any one core by other system processes. Whit

4 3

Re: [libvirt-users] How can a bridge be optimized
by Whit Blauvelt 12 Jun '12

12 Jun '12

On Tue, Jun 12, 2012 at 11:35:28AM +0400, Andrey Korolyov wrote: > Just a stupid question: did you pin guest vcpus or NIC` hw queues? > Unpinned vm may seriously affect I/O performance when running on same > core set as NIC(hwraid/fc/etc). Thanks. No question is stupid. Obviously, this shouldn't be so slow at VM I/O so I'm missing something. In my defense, there is no single coherent set of documents on this stuff, unless those are kept in a secret place. It would be a fine thing if a few of the people who know all the "obvious" stuff about libvirt-based KVM configuration would collaborate and document it fully somewhere. When I Google "pinned vcpu" all the top responses are about Xen. I run KVM. I find mention that "KVM uses the linux scheduler for distributing workload rather than actually assigning physical CPUs to VMs," at http://serverfault.com/questions/235143/can-i-provision-half-a-core-as-a-vi…. Is that wrong? Are you suggesting RAID will end up on one CPU core, and if the VM ends up on the same core there's I/O trouble? I can see that for software RAID. But we're running hardware RAID. Isn't that handled in the hardware of the RAID controller? Isn't that the point of hardware RAID, to keep it off the CPU? In any case the slow IO is true of all the VMs, so presumably they're properly spread out over CPU cores (of which there are more than VMs, and each VM is configured to take only one), which would rule out the general problem being the shared use of any one core by other system processes. Whit

1 0

[libvirt-users] ruby-libvirt connection stays open
by "Gergely Pál - night[w]" 12 Jun '12

12 Jun '12

Hi, I have a problem regarding the Ruby bindings of Libvirt. The sort summary is that the connection does not closed properly when I have retrieved a domain from it. The small sample code is the following (I use irb for easy reading of return values): The working connection closing (when I do nothing with the connection): 1.9.3p125 :001 > require 'libvirt' => true 1.9.3p125 :002 > conn = Libvirt::open('qemu+ssh://localhost/system') => #<Libvirt::Connect:0x00000001a40c30> 1.9.3p125 :003 > conn.close => nil 1.9.3p125 :004 > conn.closed? => true (Note: I know I should not use ssh for localhost, it's just a test, in general we use different remote hosts for this, so ssh is needed.) In this case there is no connection left open. But look at this code: 1.9.3p125 :001 > require 'libvirt' => true 1.9.3p125 :002 > conn = Libvirt::open('qemu+ssh://localhost/system') => #<Libvirt::Connect:0x00000001af3d80> 1.9.3p125 :003 > domain = conn.lookup_domain_by_name('gomboc18') => #<Libvirt::Domain:0x00000001adf0d8> 1.9.3p125 :004 > conn.close => nil 1.9.3p125 :005 > conn.closed? => true And i run a ps auxww there are the relevant processes: nightw 9241 0.2 0.3 88380 13256 pts/0 Sl+ 11:39 0:00 irb nightw 9248 0.0 0.0 41188 2772 pts/0 S+ 11:39 0:00 ssh localhost sh -c 'nc -q 2>&1 | grep "requires an argument" >/dev/null;if [ $? -eq 0 ] ; then CMD="nc -q 0 -U /var/run/libvirt/libvirt-sock";else CMD="nc -U /var/run/libvirt/libvirt-sock";fi;eval "$CMD";' root 9249 0.0 0.0 85752 3696 ? Ss 11:39 0:00 sshd: nightw [priv] nightw 9402 0.0 0.0 85752 1692 ? S 11:39 0:00 sshd: nightw@notty nightw 9403 0.0 0.0 4272 584 ? Ss 11:39 0:00 sh -c nc -q 2>&1 | grep "requires an argument" >/dev/null;if [ $? -eq 0 ] ; then CMD="nc -q 0 -U /var/run/libvirt/libvirt-sock";else CMD="nc -U /var/run/libvirt/libvirt-sock";fi;eval "$CMD"; nightw 9406 0.0 0.0 14136 472 ? S 11:39 0:00 nc -q 0 -U /var/run/libvirt/libvirt-sock And they remain open until i close irb (or the original ruby application on the real development system) Do you have any suggestions for solving this problem? Thanks, Pal

1 0

[libvirt-users] virsh: migration job: unexpectedly failed
by Xing.x.Gao 12 Jun '12

12 Jun '12

Hi,I am use virsh and test the migration command, on server.example.com: # rpm -qa|grep libvirt libvirt-0.9.10-21.el6.x86_64 libvirt-python-0.9.10-21.el6.x86_64 libvirt-client-0.9.10-21.el6.x86_64 # virsh version Compiled against library: libvir 0.9.10 Using library: libvir 0.9.10 Using API: QEMU 0.9.10 Running hypervisor: QEMU 0.14.1 #cat /etc/libvirt/libvirt.conf uri_aliases = [ "host1=qemu+ssh://root@host1.example.com/system", ] node1.example.com is the same version with server.example.com and on both server has the same storage pool with same path, first step,as use the local storage,i copied the vm win7's disk from server.example.com to node1.example.com's storage pool,and did a pool-refresh. 2nd step: on server.example.com i use this command to do the migration job: #virsh start win7 # virsh migrate win7 host1 --copy-storage-inc --persistent --live --unsafe when the migation process begin ,i can see the win7 appear on the node1.example.com whos status is paused,but just after a few secs,errs come out: (server.example.com) # virsh migrate win7 host1 --copy-storage-inc --persistent --live --unsafe error: operation failed: migration job: unexpectedly failed error: Reconnected to the hypervisor i've tried to solve this problem, but failed. thanks.

4 9

[libvirt-users] How to list and change domain's disk image file
by Ricardo Hillbrecht 12 Jun '12

12 Jun '12

Hi I'm writing a C program using libvirt API and need to list the storage volumes used by each domain. I'm wondering if there is a way to do this better of searching the XML strings of all domains, does anyone can suggest me something ? I also need to create a function to change the img file of a given domain. I intend to do the following: 1. virDomainGetXMLDesc (domain) 2. change the XML to other img file 3. virDomainUndefine (domain) 4. virDomainCreateXML (changed XML) Am i correct doing this ? There is a better method ? Thanks Ricardo

2 2

[libvirt-users] A RFE for libvirt
by Sebastien Douche 07 Jun '12

07 Jun '12

On Thu, Jun 7, 2012 at 12:38 AM, Eric Blake <eblake(a)redhat.com> wrote: > I personally find it easier to use virt-manager or virt-install or > gnome-boxes to create an initial guest template, then to modify that XML > for any final tweaks before using the guest in earnest. Right, but the lack of provisioning is a huge miss. Before to use a remote tool (libvirt), you must 1/ connect on the remote machine, 2/ create a xml file and 3/ define the domain. > how easy it would be to add guest templating directly into libvirt, nor > whether it would be duplicating effort that has already been coded > elsewhere. Integrating virt-install & virt-clone would be imho very good start point. -- Sebastien Douche <sdouche(a)gmail.com> Twitter: @sdouche / G+: +sdouche

1 0