- Users - libvirt List Archives

[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

[libvirt-users] question about vm management in libvirt
by Wangkai (Kevin,C) 30 Mar '13

30 Mar '13

Dear, I am a beginner of using libvirt to manage the vm, and I checked the libvirt manual. And I found the libvirt can creat, destroy, start and stop the vm, and it can manage The resource of the vm, such as storage, cpus, memory, networking and so on. And I have one doubt, Can It know the vm internal running state? Such as vm reboot, Exception, hang up, or signals received by the vm processes. I am using qemu driver. Thanks and Regards, Kevin

1 0

[libvirt-users] Support for Microsoft HyperV Server 2012
by Tejas Sumant 30 Mar '13

30 Mar '13

Hi there, Does LibVirt support Microsoft HyperV Server 2012? If not, what is plan to support the same? Thanks. -- Tejas Sumant

2 2

[libvirt-users] Snapshot without volumes temporarily
by Hubert Chiang 29 Mar '13

29 Mar '13

Hello, I want to do a snapshot which the VM has volumes. But I want to skip the volume check when I do the snapshot. (Which means I want to do a snapshot without detaching the volume) So I try to write a snapshot XML refer by http://libvirt.org/formatsnapshot.html as following: vda is my VM basic disk (qcow2) vdb is a volume (raw) ================ t1.xml =============== <domainsnapshot> <disks> <disk name='vda' snapshot='internal'> <source file='/var/lib/libvirt/VM1/disk' /> </disk> <disk name='vdb' snapshot='no' /> </disks> </domainsnapshot> ===================================== with this command: # virsh snapshot-create VM1 --xmlfile t1.xml and get the message: # error: argument unsupported: unable to handle disk requests in snapshot My environment is Libvirt 0.9.8, Qemu 1.0.0 on Ubuntu12.04.02 I try to do this action before at libvirt 0.9.2 QEMU 0.14.1 on Ubuntu 11.10 before with following commands, It's success. Step1. I copy the XML from /etc/libvirt/qemu/VM1.xml to VM1.xml.backup Step2. I edit the VM1.xml to remove the disk tag which is create by volume Step3. do virsh command: virsh snapshot-create VM1 Step4. Move VM1.xml.backup to VM1.xml It's success on libvirt 0.9.2. But on 0.9.8 It doesn't work. Because libvirt will check the setups from memory not the XML file in /etc/libvirt/qemu/ What could I do with libvirt 0.9.8? Thanks in advance.

2 2

[libvirt-users] virsh migrate ---no route to host
by Hari Pyla 28 Mar '13

28 Mar '13

Hi, I am trying to migrate a guest domain from one node to another. I've tried several options of the virsh migrate command, but in vain. It seems to be a networking issue. I wanted to make sure that my setup is correct and that I am not missing anything. I've issued the below command on the source node (n0): [user@n0 ~]$ virsh --connect qemu:///system migrate --verbose Fedora-17-x86_64-1 qemu+ssh://n1/system error: unable to connect to server at 'n1:49156': No route to host I was wondering if you have any ideas on how to fix this issue. My current setup/configuration: 1. virsh version : 0.10.2 2. The user has all the necessary access control setup. I've referenced http://libvirt.org/auth.html I am able to locally and remotely connect to both the nodes. I am also able to start, stop and perform other actions on guest domains on local and remote nodes, e.g., /* access destination node */ [user@n0 ~]$ virsh --connect qemu+ssh://n1/system list --all Id Name State ---------------------------------------------------- /* access local node */ [user@n0 ~]$ virsh --connect qemu:///system list --all Id Name State ---------------------------------------------------- 2 Fedora-17-x86_64-1 running - Fedora-18-x86_64-DVD shut off 3. ssh keys are configured for password less access to both the nodes 4. I've opened ports 49152-49215 on the destination node i.e., n1. please see below. [user@n1 images]$ sudo iptables -L [sudo] password for user: Chain INPUT (policy ACCEPT) target prot opt source destination ACCEPT udp -- anywhere anywhere udp dpt:domain ACCEPT tcp -- anywhere anywhere tcp dpt:domain ACCEPT udp -- anywhere anywhere udp dpt:bootps ACCEPT tcp -- anywhere anywhere tcp dpt:bootps ACCEPT all -- anywhere anywhere state RELATED,ESTABLISHED ACCEPT icmp -- anywhere anywhere ACCEPT all -- anywhere anywhere ACCEPT tcp -- anywhere anywhere state NEW tcp dpt:ssh REJECT all -- anywhere anywhere reject-with icmp-host-prohibited ACCEPT tcp -- anywhere anywhere tcp dpts:49152:49215 ACCEPT tcp -- anywhere anywhere tcp spts:49152:49215 Chain FORWARD (policy ACCEPT) target prot opt source destination ACCEPT all -- anywhere 192.168.122.0/24 state RELATED,ESTABLISHED ACCEPT all -- 192.168.122.0/24 anywhere ACCEPT all -- anywhere anywhere REJECT all -- anywhere anywhere reject-with icmp-port-unreachable REJECT all -- anywhere anywhere reject-with icmp-port-unreachable REJECT all -- anywhere anywhere reject-with icmp-host-prohibited Chain OUTPUT (policy ACCEPT) target prot opt source destination 5. I've restarted the libvirtd after updating the iptables. I don't think that is necessary, but did it anyway. I was wondering if I am missing something. Any help is greatly appreciated. Thanks, --Hari

1 0

[libvirt-users] virsh migrate --no route to host
by Hari Pyla 28 Mar '13

28 Mar '13

Hi, I am trying to migrate a guest domain from one node to another. I've tried several options of the virsh migrate command, but in vain. It seems to be a networking issue. I wanted to make sure that my setup is correct and that I am not missing anything. I've issued the below command on the source node (n0): [user@n0 ~]$ virsh --connect qemu:///system migrate --verbose Fedora-17-x86_64-1 qemu+ssh://n1/system error: unable to connect to server at 'n1:49156': No route to host I was wondering if you have any ideas on how to fix this issue. My current setup/configuration: 1. virsh version : 0.10.2 2. The user has all the necessary access control setup. I've referenced http://libvirt.org/auth.html I am able to locally and remotely connect to both the nodes. I am also able to start, stop and perform other actions on guest domains on local and remote nodes, e.g., /* access destination node */ [user@n0 ~]$ virsh --connect qemu+ssh://n1/system list --all Id Name State ---------------------------------------------------- /* access local node */ [user@n0 ~]$ virsh --connect qemu:///system list --all Id Name State ---------------------------------------------------- 2 Fedora-17-x86_64-1 running - Fedora-18-x86_64-DVD shut off 3. ssh keys are configured for password less access on both the nodes 4. I've opened ports 49152-49215 on the destination node i.e., n1. please see below. [user@n1 images]$ sudo iptables -L [sudo] password for user: Chain INPUT (policy ACCEPT) target prot opt source destination ACCEPT udp -- anywhere anywhere udp dpt:domain ACCEPT tcp -- anywhere anywhere tcp dpt:domain ACCEPT udp -- anywhere anywhere udp dpt:bootps ACCEPT tcp -- anywhere anywhere tcp dpt:bootps ACCEPT all -- anywhere anywhere state RELATED,ESTABLISHED ACCEPT icmp -- anywhere anywhere ACCEPT all -- anywhere anywhere ACCEPT tcp -- anywhere anywhere state NEW tcp dpt:ssh REJECT all -- anywhere anywhere reject-with icmp-host-prohibited ACCEPT tcp -- anywhere anywhere tcp dpts:49152:49215 ACCEPT tcp -- anywhere anywhere tcp spts:49152:49215 Chain FORWARD (policy ACCEPT) target prot opt source destination ACCEPT all -- anywhere 192.168.122.0/24 state RELATED,ESTABLISHED ACCEPT all -- 192.168.122.0/24 anywhere ACCEPT all -- anywhere anywhere REJECT all -- anywhere anywhere reject-with icmp-port-unreachable REJECT all -- anywhere anywhere reject-with icmp-port-unreachable REJECT all -- anywhere anywhere reject-with icmp-host-prohibited Chain OUTPUT (policy ACCEPT) target prot opt source destination 5. I've restarted the libvirtd after updating the iptables. I dont think that is necessary, but did it anyway. I was wondering if I am missing something. Any help is greatly appreciated. Thanks, --Hari

1 0

[libvirt-users] netfilter+libvirt=(smth got broken?)
by Nikolai Zhubr 28 Mar '13

28 Mar '13

Hello, I'm having problem setting up filtering traffic for a virtual machine managed by libvirt. Strange thing is, such a setup has been working fine for me on an older version of distro (namely, opensuse 11.3 w/updates, kernel 2.6.34, libvirt 0.8.8) but refused to work on shiny new opensuse 12.4 (kernel 3.7.10, libvirt 1.0.2). The definition of filter in question is pretty simple: <filter name='some-filt' chain='ipv4'> <rule action='accept' direction='in'> <tcp dstportstart='110'/> </rule> <rule action='drop' direction='inout'> <all/> </rule> </filter> So basically it should allow incoming connections to the specified port number and nothing else. After activating this filter on a box in question, connections to 110 started to fail (timeout). Examining iptables rules manually and comparing them to the rules from my old box did not reveal anything suspicious to me. However, through just pure guesswork, I managed to ocasionally "fix" the problem by manually editing 3 relevant rules as follows: --A FI-vnet0 -p tcp -m tcp --sport 110 -m conntrack --ctstate ESTABLISHED -m conntrack --ctdir ORIGINAL -j RETURN +-A FI-vnet0 -p tcp -m tcp --sport 110 -m conntrack --ctstate ESTABLISHED -m conntrack --ctdir REPLY -j RETURN --A FO-vnet0 -p tcp -m tcp --dport 110 -m conntrack --ctstate NEW,ESTABLISHED -m conntrack --ctdir REPLY -j ACCEPT +-A FO-vnet0 -p tcp -m tcp --dport 110 -m conntrack --ctstate NEW,ESTABLISHED -m conntrack --ctdir ORIGINAL -j ACCEPT --A HI-vnet0 -p tcp -m tcp --sport 110 -m conntrack --ctstate ESTABLISHED -m conntrack --ctdir ORIGINAL -j RETURN +-A HI-vnet0 -p tcp -m tcp --sport 110 -m conntrack --ctstate ESTABLISHED -m conntrack --ctdir REPLY -j RETURN So essentially, just manually swtiching "--ctdir" values to their opposites makes the filter allow connections to port 110. I've also verified that the filter still blocks unwanted connections originating from port 110 from VM exactly as it should: (in VM): netcat -v -v -n -p 110 192.168.122.1 22 (UNKNOWN) [192.168.122.1] 22 (?) : Connection timed out set 0, rcvd 0 I've then compared /proc/net/nf_conntrack on both (old and new) boxes. They look roughly the same, nothing suspicious. This all looks to me as if "--ctdir" argument somehow magically changed its meaning to the opposite, but this just cannot be! I'm out of ideas and looking for insights. Any hints appreciated quite a lot. Thank you. Nikolai

4 12