- Users - libvirt List Archives

[libvirt-users] Forwarding incoming connections from an secondary external interface with dhcp
by Bhima Pandava 28 Apr '14

28 Apr '14

Hi everyone, I have a server (ubuntu 14.04 lts) running KVM with a few guests (each with static IP addresses) which I manage using libvirt. This machine has a wired interface (br0 / p2p1) connecting to my internal network and a wireless interface using dhcp (ppp0) as a secondary direct connection to the internet. My goal is route all administration traffic for both the host and guests through the wired interface (br0); while routing the service traffic, and _only_ this traffic, through the ppp0 interface. For example: One vm is a webserver and I wish to route all traffic on ports 80 & 443 from ppp0 to that vm. Another vm is a mailserver and I wish to route all the imap & pop traffic from ppp0 to that vm. At the same time, I wish to be able to ssh in through the wired interface (p2p1) and route all system maintenance traffic through the wired interface as well. Additionally I wish to drop/reject incoming traffic on ppp0 which is not on these specified ports. Reading the libvirt documentation, I have found a section which very nearly describes my needs: http://wiki.libvirt.org/page/Networking#Forwarding_Incoming_Connections Unfortunately, I am unsure how to correcting modify the given examples to meet my specific needs. To begin with, the examples assume a static external IP, while mine is dynamic. I think I would like to refer to interface by name (ppp0) or perhaps create a secondary bridge (br1) with a static IP, the ppp0 included with "bridge_prts ppp0" command, and then refer to the static IP of the new bridge (br1). Next, I don't see the correct mechanism which I could configure such that all traffic coming in ppp0 which was not on the specified ports is rejected/dropped and all administrative traffic, originating from both the host and guests, is routed through the wired interface. I am assuming that this also should be done through libvirt/virsh and not directly through iptables. If it matters, the wired interface connects to a router running shorewall (and thus the server is in the "dmz" zone). The server itself is not running shorewall or any other similar things. The router is managing the primary connection to internet, which is distinct from the secondary wireless interface on the server, it is connecting to a different ISP. If anyone could give me some advice or point me to some additional or more relevant documentation/examples I would really, really appreciate it. Thanks!

1 0

[libvirt-users] Ability of a VM to detect Libvirt events.
by Taimur Al Said 26 Apr '14

26 Apr '14

Hi there, Let's assume a libvirt event occurred on a VM, i.e an event like pause VM or reboot VM or any other libvirt event. Can the VM determine that such event was originated by libvirt? ---or in a more general term by the hypervisor? If yes, how can this be done in theory? Thanks in advance, Regards, Taimur

2 1

[libvirt-users] Deriving VMs from a reference image via scripts
by Philip Prindeville 25 Apr '14

25 Apr '14

Hi. I have a reference .vmdk (and .xml) reference VM. I need to instantiate multiple new VM’s from this, but changing various parameters: name, MAC address, network connections, cores, memory size, disk size and backing store, etc. and then start these new VM’s… all via ‘virsh’ et al from a script. What are the steps to do this? I was looking at the command help and some tutorials but apparently not the right ones. Thanks, -Philip

2 1

[libvirt-users] Using IVSHMEM with Libvirt
by Qasim Maqbool 24 Apr '14

24 Apr '14

Hi, I am trying to create a VM with the IVSHMEM feature, and I specify the device using qemu command line argument in my XML by using the pass through tag as follows: <qemu:commandline> <qemu:arg value='-device ivshmem,size=2048M,shm=fd:/mnt/huge/map_1:0x0:0x40000000:/dev/zero:0x0:0x3fffc000:/var/run/.ivshmem_metadata_config_2:0x0:0x4000'/> </qemu:commandline> This seemingly does not work as I get this error: Internal error process exited while connecting to monitor: qemu-system-x86_64: -device ivshmem,size=2048M,shm=fd:/mnt/huge/map_1:0x0:0x40000000:/dev/zero:0x0:0x3fffc000:/var/run/.ivshmem_metadata_config_2:0x0:0x4000: invalid option Can anyone point out what I am doing wrong here? Thanks. Qasim

2 1

[libvirt-users] Libvirt qemu permission denied error
by Asadullah Hussain 23 Apr '14

23 Apr '14

Hello, I am trying run a simple VM (qemu1.4.2) on RHEL 6.5 through libvirt (0.10). virsh create throws the error: *error: Failed to create domain from vm1.xml* *error: internal error process exited while connecting to monitor:* The libvirt log for VM shows: *libvir: error : cannot execute binary /usr/bin/qemu-system-x86_64: Permission denied* I am running VM as root and all disk images/ISOs required for VM are read/write/executable. -- Asadullah Hussain

2 3

[libvirt-users] TLS and intermediate CA
by Nathaniel Cook 22 Apr '14

22 Apr '14

I have been trying to get set of libvirtd system up and running. My PKI infrastructure involves a root CA and several intermediate CAs. I am trying to get the machines to trust each other across the different intermediate CAs. This is what I have so far: Libvirtd is starting and listening on tls port 16514 I have configured client/server certs/keys and it seems to be using all of these correctly. I have also configured the cacert.pem file (which has two certs in the chain). I have confirmed (recompiling with various debug statements) that the gnutls libraries are successfully loading both certs from the cacert.pem file. When I try to connect with openssl s_client -connect <host>:16514 I get something similar to this: --- Certificate chain 0 s:/CN=kvm999.example.com i:/C=US/ST=Utah/O=Qualtrics/OU=SRE/CN=intca1.example.com --- Server certificate -----BEGIN CERTIFICATE----- ... omitted for brevity -----END CERTIFICATE----- subject=/CN=kvm999.example.com issuer=/C=US/ST=Utah/O=Qualtrics/OU=SRE/CN=intca1.example.com --- Acceptable client certificate CA names /C=US/ST=Utah/O=Qualtrics/OU=SRE/CN=intca1.example.com /C=US/ST=Utah/O=Qualtrics/OU=SRE/CN=rootca.example.com --- The "Server certificate" and "Acceptable client certificate CA names" look right. The problem is that the certificate chain is just the single server cert and does not include the intermediate cert or root cert. As a result clients from other intermediate CAs fail to verify the libvirtd process. I have tried libvirtd 0.10.2 and 1.2.1 both on CentOS 6. Thanks. -- -Nathaniel Cook

2 4

[libvirt-users] Live snapshot merging (qemu 2.0)
by Thomas Stein 22 Apr '14

22 Apr '14

Hello. The Changelog of qemu-2.0.0 mentioned "Live snapshot merging". Someone has an idea what could be ment by this? I'm asking because i'm still struggling with finding a reliable backup solution for running kvm machines. Blockcopy is my current solution. best regards Thomas

2 2

[libvirt-users] "virt-install" source location boot disk
by Jay Vyas 22 Apr '14

22 Apr '14

Hi virt, im stumped... any help would be appreciated. I normally create my VMs like this: base="http://mirror.pnl.gov/fedora/linux/releases/20/Fedora/x86_64/os/" sudo virt-install --hvm --name $vm_name$i --ram 4000 \ --disk path=/VirtualMachines/$vm_name$i,size=30 \ --location $base -x "ks=http://xxx.os21.ks$kx" HOWEVER... I'm finding that my VMs dont boot after i restart the Host. My question is: Where is virt-install writing the boot disk to? I get a error message that the "boot failed" because the hard disk, is not a bootable disk. But I can't imagine where else to boot from ? Because I point my --location of the original boot to the source code repo itself.... ANY help would be appreciated thanks!

2 5

[libvirt-users] routed subnet problem
by Patric Buskas 22 Apr '14

22 Apr '14

Hi, I'm new to libvirt so please bear with me. I am trying to set up a routed subnet on my laptop with libvirt, hoping it to be able to communicate with all servers in my nw. My laptop is on a subnet, 192.168.2.0/24, with a static ip, 192.168.2.27. I have "more or less" followed this guide for routed networks, https://www.berrange.com/posts/2009/12/13/routed-subnets-without-nat-for-li… The "more or less" part is that I have a Asus RT-N56U router with the static route set as in the guide and I'm not running wifi. With this setup I can reach, within the 192.168.200.0/24, all guests within 192.168.200.0/24, internet and the laptop but I can not reach any server on the laptop subnet. traceroute from guests to servers in the laptop subnet is only giving "stars" infinitely nmap from guests to services in the laptop subnet is saying "filtered" Obviously I've done something wrong but I can't find out what. I also may have misunderstood the meaning of routed network within libvirt. Please give me some advise to solve this problem. Some facts: The laptop is running Ubuntu 13.04 The net-dumpxml https://gist.github.com/kakbit/11103807 Routing on laptop https://gist.github.com/kakbit/11103832 iptables on laptop https://gist.github.com/kakbit/11103947

2 1

[libvirt-users] Guest Virtual Bios in XEN
by Mui Le Van 21 Apr '14

21 Apr '14

Hi, I am using Libvirt with XEN hypervisor to manage the VM. I see there are 3 kinds of Virtual BIOS for XEN Guest VM, they are ROMBIOS,SEABIOS, OVMF ( is disabled by default as XEN Supporters said ). But I don't know how to pass the BIOS options to the Guest from XML file. Could you please suggest how to change the Virtual BIOS for the Guest from XML file in Libvirt or give some clues to do that? Below is the link that I asked XEN support: http://osdir.com/ml/general/2014-04/msg32102.html Thank for your time! Mui Le,

1 0