April 2013 - Users - libvirt List Archives

[libvirt-users] Cannot recv data: Connection reset by peer
by Guillaume Thouvenin 03 Apr '13

03 Apr '13

Hello, I'm trying to install and use libvirt 1.0.4 on an Ubuntu 12.04.2. I compiled libvirt by doing: $ ./configure --with-selinux=no --with-gnutls $ make $ sudo make install Everything works fine and are installed under /usr/local Then I started the libvirtd manually and try to start a linux container that has the following configuration: $ virsh -c lxc:/// dumpxml lxcvm1 <domain type='lxc'> <name>lxcvm1</name> <uuid>168edaae-e3b5-2d06-7c65-9aa444a6f234</uuid> <memory unit='KiB'>2097152</memory> <currentMemory unit='KiB'>2097152</currentMemory> <vcpu placement='static'>1</vcpu> <os> <type arch='x86_64'>exe</type> <init>/bin/sh</init> </os> <clock offset='utc'/> <on_poweroff>destroy</on_poweroff> <on_reboot>restart</on_reboot> <on_crash>destroy</on_crash> <devices> <emulator>/usr/local/libexec/libvirt_lxc</emulator> <filesystem type='mount' accessmode='passthrough'> <source dir='/home/thouveng/lxc_dir/lxcvm1/rootfs'/> <target dir='/'/> </filesystem> <console type='pty'> <target type='lxc' port='0'/> </console> </devices> </domain> When I start the container I have the following error: $ virsh -c lxc:/// start lxcvm1 error: Failed to start domain lxcvm1 error: internal error guest failed to start: PATH=/bin:/sbin TERM=linux container=lxc-libvirt container_uuid=168edaae-e3b5-2d06-7c65-9aa444a6f234 LIBVIRT_LXC_UUID=168edaae-e3b5-2d06-7c65-9aa444a6f234 LIBVIRT_LXC_NAME=lxcvm1 /bin/sh error receiving signal from container: Input/output error The libvirtd is started with info option so I can see: 2013-04-03 08:49:20.167+0000: 17046: error : virNetSocketReadWire:1362 : Cannot recv data: Connection reset by peer 2013-04-03 08:49:20.267+0000: 17051: error : virLXCProcessStart:1248 : internal error guest failed to start: PATH=/bin:/sbin TERM=linux container=lxc-libvirt container_uuid=168edaae-e3b5-2d06-7c65-9aa444a6f234 LIBVIRT_LXC_UUID=168edaae-e3b5-2d06-7c65-9aa444a6f234 LIBVIRT_LXC_NAME=lxcvm1 /bin/sh error receiving signal from container: Input/output error In libvirtd.conf I have: auth_unix_ro = "none" auth_unix_rw = "none" auth_tcp = "none" auth_tls = "none" And in libvirt.conf everything is commented. I have this error as normal user (in libvirtd group) and also as root. Any hints to go further in the debugging are welcome. Regards, Guillaume

2 2

[libvirt-users] migrate qemu domains
by Yin Olivia-R63875 03 Apr '13

03 Apr '13

Hi, I tried to migrate qemu domains between same two Freescale PPC platforms. 1) Migrate test domain from Host1(10.193.20.109) to Host2(10.193.20.181). root@ppc-host1:~# virsh migrate test qemu+tls://10.193.20.181/system or root@ppc-host2:~# virsh -c qemu+tls://10.193.20.109/system migrate test qemu:///system I waited for a while and finally got error message as below: error: Unable to read from monitor: Connection reset by peer Exactly, when waiting (I guess during migration), the migrated domain is pasued on both source and destination nodes. root@ppc-host1:~# virsh -c qemu:///system list --all Id Name State ---------------------------------------------------- 5 test paused And meanwhile, the destination domain is also paused when waiting root@ppc-host2:~# virsh -c qemu:///system list --all Id Name State ---------------------------------------------------- 2 test paused Finally, the migrated domain on source node is shut off. root@ppc-host1:~# virsh -c qemu:///system list --all Id Name State ---------------------------------------------------- - test shut off But can't find any domain on destination node any longer. root@ppc-host2:~# virsh -c qemu:///system list --all Id Name State ---------------------------------------------------- As I know qemu-system-ppc doesn't support live migration on Freescale PPC platform yet. My question is whether it depends on qemu implementation to migrate qemu hypervisor with libvirt? >From the above experiment, it seemed that migration happened but not succeeded finally. Best Regards, Olivia

3 4

[libvirt-users] virt-manager connect remote KVM host and graphic control guest problem
by cochen 03 Apr '13

03 Apr '13

Hi developer, i got a problem and need your help!(my english not very good) My KVM host in IDC, and there are some restrict in there lead to me connect to KVM host by port 22 or 16509 impossible. In generally If i want login my KVM host i must login in to a "stepping stones" first, then login to that KVM host.( Now assume that the KVM host ip is 1.1.1.1 and the "stepping stones" ip is 2.2.2.2). So i use the "portforwarding" (i used the manner of "qemu+tcp" to connect KVM host, the default port is 16509), I visit the port 16509 in "stepping stones" machine 1.1.1.1 and then 1.1.1.1 port forwarding to 2.2.2.2:16509. And then my virt-manager connect KVM host successful！ But my trouble is comming too. I can't use virt-manager to control my guest by graphic. I "portforwarding" the port 5911(i had already defined my guest vnc port), but is unuseful. I used command "netstat -antl" in KVM host found that : tcp 0 0 127.0.0.1:5911 0.0.0.0:* LISTEN tcp 0 0 1.1.1.1:16509 2.2.2.2:63325 ESTABLISHED tcp 0 0 0.0.0.0:16509 0.0.0.0:* LISTEN As you see, 5911 listen into localhost(127.0.0.1), but 16509 listen into the 1.1.1.1 and 0.0.0.0. So I think if 5911 listen 0.0.0.0 or 1.1.1.1 everything is ok. Here is my virt-manager.log: [Thu, 28 Mar 2013 00:57:16 virt-manager 3242] DEBUG (engine:471) window counter incremented to 2 [Thu, 28 Mar 2013 00:57:16 virt-manager 3242] DEBUG (console:1078) Starting connect process for proto=vnc trans=tcp connhost=2.2.2.2 connuser=root connport=None gaddr=127.0.0.1 gport=5911 gsocket=None [Thu, 28 Mar 2013 00:57:16 virt-manager 3242] DEBUG (console:374) VNC connecting to 2.2.2.2:5911 [Thu, 28 Mar 2013 00:57:18 virt-manager 3242] DEBUG (console:961) Viewer disconnected How can i use virt-manager to control my guest by graphic? thank you very much!!!

2 1

[libvirt-users] How to give access to /dev/tty
by Guillaume Thouvenin 02 Apr '13

02 Apr '13

Hello, First I apologize because I sent this email on the lxc-users mailing by error. So sorry for the double message. I'm trying to run an X server inside a container. I was able to achieve this with lxc-tools (and that is why I mixed myself between the 2 lists) by setting the option lxc.tty = 4 that allows access to 4 tty. Now I want to use libvirt because my final goal is to create a container with GPU capabilities from Openstack (that uses libvirt). When I start an X server with "startx -- :1" in a container I have the error: Fatal server error: xf86OpenConsole: Cannot open /dev/tty0 (No such file or directory) I also tried with :2 and :3 but I still have the same error (except that the name of the log file changed between Xorg.{1,2,3}.log. I copied the xml file of my container at the end of the email. Thus my question is how can I give an access to /dev/tty0 to the container? Or maybe it is not the right way to start an X server in the container. Best regards, Guillaume <domain type='lxc' id='28247'> <name>lxcvm1</name> <uuid>d4014dfc-f4db-5791-c0d2-817547921ce1</uuid> <memory unit='KiB'>1048576</memory> <currentMemory unit='KiB'>1048576</currentMemory> <vcpu placement='static'>1</vcpu> <os> <type arch='x86_64'>exe</type> <init>/sbin/init</init> </os> <clock offset='utc'/> <on_poweroff>destroy</on_poweroff> <on_reboot>restart</on_reboot> <on_crash>restart</on_crash> <devices> <emulator>/usr/lib/libvirt/libvirt_lxc</emulator> <filesystem type='mount' accessmode='passthrough'> <source dir='/home/thouveng/virtualization/lxc/lxcvm1/rootfs'/> <target dir='/'/> </filesystem> <interface type='network'> <mac address='00:16:3e:3e:06:4d'/> <source network='default'/> <target dev='veth0'/> </interface> <console type='pty' tty='/dev/pts/2'> <source path='/dev/pts/2'/> <target type='lxc' port='0'/> <alias name='console0'/> </console> <console type='pty'> <source path='/dev/pts/6'/> <target type='lxc' port='1'/> <alias name='console1'/> </console> <console type='pty'> <source path='/dev/pts/7'/> <target type='lxc' port='2'/> <alias name='console2'/> </console> </devices> <seclabel type='none'/> </domain>

4 7

[libvirt-users] problem in view vm created by virsh create
by 高松 02 Apr '13

02 Apr '13

<domain type='kvm'> <name>eucalyptus-windows</name> <os> <type>hvm</type> <boot dev='cdrom'/> </os> <features> <acpi/> </features> <memory>524288</memory> <vcpu>1</vcpu> <devices> <emulator>/usr/libexec/qemu-kvm</emulator> <disk type='file'> <source file='/root/windows.server_2003_sp2.img'/> <target dev='hda'/> </disk> <disk type='file' device='disk'> <source file='/root/secondary.img'/> <target dev='vda' bus='virtio'/> </disk> <disk type='file' device='floppy'> <source file='/root/floppy.img'/> <target dev='fda'/> </disk> <disk type='file' device='cdrom'> <source file='/root/windows_server_2003_sp2.iso'/> <target dev='hdc'/> <readonly/> </disk> <interface type='bridge'> <source bridge='eucabr899'/> <model type='rtl8139'/> </interface>  <graphics type='vnc' port='-1' listen='0.0.0.0' passwd='123456'/> </devices> </domain> ========================================================= I issue the command of "virsh create libvirt-kvm-windows.xml" to create a vm, and it is successful.Above is the content,but when I try to connect to the VM both try vinagre and virt-manager,what I got is only black screen,I can't understand and please help!

2 1

[libvirt-users] nparams in virNodeGetCPUStats and CPU utilization
by Hari Pyla 02 Apr '13

02 Apr '13

Hi, I am trying to get the CPU utilization of a node and I am using virNodeGetCPUStats(...) Here is my code snippet. The error checking is omitted for brevity. /* get the number of params */ retval = virNodeGetCPUStats (conn, VIR_NODE_CPU_STATS_ALL_CPUS,\ NULL, &nparams, 0); printf ("nparams: %d\n", nparams); /* allocate space for the params */ params = (virNodeCPUStatsPtr *) malloc (sizeof (virNodeCPUStats) * nparams ); /* get the values */ retval = virNodeGetCPUStats (conn, VIR_NODE_CPU_STATS_ALL_CPUS,\ params, &nparams, 0); for ( i = 0 ; i < nparams ; i++) { printf ("params[%d].field: %s \tparams[%d].value: %llu\n", i, params[i].field, i, params[i].value); } output: nparams: 4 params[0].field: kernel params[0].value: 77930000000 params[1].field: user params[1].value: 568270000000 params[2].field: idle params[2].value: 471429230000000 params[3].field: iowait params[3].value: 67100000000 The documentation at http://libvirt.org/html/libvirt-libvirt.html on the virNodeGetCPUStats(...) API says that nparams is dynamic. Does that mean that not all params are available for a node? I was hoping that nparams would be 5 and that params[4] would contain the utilization (VIR_NODE_CPU_STATS_UTILIZATION). If the answer to the above is 'yes'. Then, I was wondering as to what would be the correct way to calculate the CPU utilization of a node?. Thanks, --Hari

1 0

[libvirt-users] XML error: missing security model on virsh migrate
by Hari Pyla 02 Apr '13

02 Apr '13

Hi, I am trying to migrate a VM from one node to another and I get the following error message. [user@n0 ~]$ virsh --c qemu:///system migrate --verbose Fedora-17-x86_64-1 qemu+ssh://n1/system error: XML error: missing security model when using multiple labels On both the src and dest nodes, I've disabled SELinux and I have the following libvirt version installed $virsh --connect qemu:///system version --daemon Compiled against library: libvirt 0.10.2 Using library: libvirt 0.10.2 Using API: QEMU 0.10.2 Running hypervisor: QEMU 0.12.1 Running against daemon: 0.10.2 Any help is greatly appreciated. Thanks, --Hari

3 4