- Users - libvirt List Archives

[libvirt-users] Questions on libvirt hook scripts
by Whit Blauvelt 11 Jul '11

11 Jul '11

Hi, Reading http://libvirt.org/hooks.html I'm wondering of the following implication is accurate: Under "QEMU guest migration" it says At the beginning of the migration, the qemu hook script on the _destination_ host is executed with the "start" operation. Above that, under "Specifics" Before a QEMU guest is started, the qemu hook script is called in two locations; if either location fails, the guest is not started. The first location, since 0.9.0, is before libvirt performs any resource labeling.... This is called as: /etc/libvirt/hooks/qemu guest_name prepare begin - The second location, available Since 0.8.0, occurs after libvirt has finished labeling all resources, but has not yet started the guest, called as: /etc/libvirt/hooks/qemu guest_name start begin - Is it accurate that under 0.9.x, starting a VM will run the qemu hook script as both "prepare" and "start," but migrating a VM will only run the qemu hook script as "start"? Or is the description incomplete with libvirt also running "prepare" on migration? Another question, under "Calling libvirt functions from within a hook script" it says DO NOT DO THIS! A hook script must not call back into libvirt, as the libvirt daemon is already waiting for the script to exit. A deadlock is likely to occur. That makes obvious sense. But does it apply to calling libvirt on a second system? For example, a script might check the status of VMs on a second system as a condition of starting a VM on first, using virsh. My guess is that shouldn't deadlock. But then since during migration "the qemu hook script on the _destination_ host is executed," if the destination hook script were in turn to use virsh to check the _source_ host, then in a roundabout way a libvirt function has been called from within the hook script. I can't quite picture whether this particular instance would be dangerous. If it's accurate that the _destination_ hook script is only called with "start" and not with "prepare," the second question matters less. The checks on the second system that I want to do can be limited to the "prepare" invocation, and so never loop back. On the other hand, if the destination hook script _is_ called with "prepare" during migration, it would be useful to know how to tell if it's being called as part of a migration. TIA, Whit

1 0

[libvirt-users] NPIV + KVM
by Eric Ross 11 Jul '11

11 Jul '11

Is this possible: I would like to have a VM use a FC LUN exported to a virtual wwn (using NPIV) as its primary disk. I was thinking there might be a mechanism to say VM, boot using a LUN with this vwwn as your primary disk? Thanks in advance. -- --------------------------------------------------------------------- Eric Ross HPC Systems Administrator Argonne National Laboratory Leadership Computing Facility ---------------------------------------------------------------------

2 1

[libvirt-users] Creating a CENTOS VM
by Dale Amon 11 Jul '11

11 Jul '11

I used the following command: virt-install --connect qemu:///system -n server -r 512 --vcpus=1 -f kvm/server.cyberteams.com/root.qcow2 -s 12 -c iso/CentOS-5.6-x86_64-bin-DVD-1of2.iso --vnc --noautoconsole --os-type linux --accelerate --network=bridge:br0 --hvm which with a BlueOnyx CD some months ago worked fine. However when I connected with virt-viewer, after going through the TEST function it then claimed it could not find the CD. Is CENTOS a bit strange in some way?

1 1

[libvirt-users] Changing path of VM configuration files
by Jatin Kumar 08 Jul '11

08 Jul '11

Hello, I am using shared storage and i want to keep all the configuration files of vms of different hosts in one single location (on shared storage). Is there a way i can change the default path=/etc/libvirt/qemu/<VMName>.xml to something else. I looked into libvirtd.conf and qemu.conf but could not find any such thing. -- Jatin

3 2

[libvirt-users] KVM and <interface type='ethernet'>
by Maximilian Wilhelm 07 Jul '11

07 Jul '11

Hi! I'm trying to set up a VM with "just an tap device" with the following config (basicly inherited from 'virsh domxml-from-native qemu-argv'): <domain type='kvm'> [...] <interface type='ethernet'> <mac address='52:54:1f:ac:08:51'/> <target dev='foo_eth0'/> <model type='virtio'/> </interface> [...] </devices> </domain> This sadly failes with: | error: Failed to start domain foo | error: internal error process exited while connecting to monitor: could not configure /dev/net/tun: Device or resource busy I'm wondering why this fails as there shouldn't anything be done with the tap device besides setting it up. (The basic goal would be to set up a VM with a tap device, put an v4 and v6 IP on it and let OSPF do the rest. As the <script> parameter will not work with dropped privileges I wrote a wrapper for the kvm binary which takes care of this via sudo. This part works, but I fail to get the VM started with <interface type='ethernet'>) Any ideas? Thanks Max -- <@uschebit> Das erklärt die Sitte. Das ist so ähnlich wie mit dem Verbeugen zur Begrüßung in Japan. Das dient auch nur dazu, durch Begutachten der Schuhqualität den sozialen Rang festzustellen.

1 0

[libvirt-users] Is arping useful if a VM is brought up on a different (bridged) interface/machine?
by Whit Blauvelt 07 Jul '11

07 Jul '11

Hi, Using libvirt/qemu/kvm VMs, if a VM has stopped on one host and then the identical VM is started afresh on a different host, with the exact same (bridged) VM IP, would there be a point in having the VM run arping (or arping2) on startup to help it be quickly found on the LAN? Or is this built in at some level already? Or does it not apply here? Thanks, Whit

1 0

[libvirt-users] SSH console for qemu
by Nikita A Menkovich 07 Jul '11

07 Jul '11

Hello, I'm using qemu with libvirt and I want to create SSH emergency console to linux/unix guests, same way as it is realized in qemu VNC. Maybe someone could help me to determine a place to dive in this question. I think the best way to make it - extend qemu API. But if this could be made with libvirt - this will be also fine. -- Nikita A Menkovich http://libc6.org/ JID: menkovich(a)gmail.com

3 5

[libvirt-users] Guest-networking: Two subnets = Guests cannot connect to the other subnet
by mailinglists＠florian-kaiser.net 07 Jul '11

07 Jul '11

Hello list, We have KVM setup using libvirt (0.9.0) as configuration device on a Debian Squeeze Host. Networking is done in routed bridge mode and we have two different subnets, as well as one main ip. Subnets get routed to the main ip (this is done by our provider). Network on hosts are done like this: <network> <name>subnet_1</name> <uuid>7c60990d-ab31-c4a5-cd2b-5ce4c7c61d38</uuid> <forward dev='eth0' mode='route'/> <bridge name='virtbridge1' stp='off' delay='0' /> <mac address='52:54:00:4D:83:2A'/> <ip address='xx.xx.240.121' netmask='255.255.255.248'> </ip> </network> <network> <name>subnet_2</name> <uuid>7c60990d-ab31-c4a5-cd2b-5ce4c7c61d38</uuid> <forward dev='eth0' mode='route'/> <bridge name='virtbridge1' stp='off' delay='0' /> <mac address='52:54:00:4D:83:2A'/> <ip address='xx.xx.240.200' netmask='255.255.255.248'> </ip> </network> Routing on host Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use Iface xx.xx.122.33 0.0.0.0 255.255.255.255 UH 0 0 0 eth0 xx.xx.240.120 0.0.0.0 255.255.255.248 U 0 0 0 virtbridge1 xx.xx.239.200 0.0.0.0 255.255.255.248 U 0 0 0 virtbridge2 0.0.0.0 xx.xx.122.33 0.0.0.0 UG 0 0 0 eth0 Guests get this: <interface type='network'> <mac address='de:ad:b6:83:85:5b'/> <source network='subnet_1'/> <model type='virtio'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x03' function='0x0'/> </interface> Guests put this into interfaces: iface eth0 inet static address xx.xx.239.202 netmask 255.255.255.248 gateway xx.xx.239.201 Guest routing Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use Iface xx.xx.239.200 0.0.0.0 255.255.255.248 U 0 0 0 eth0 0.0.0.0 xx.xx.239.201 0.0.0.0 UG 0 0 0 eth0 Everything seemed to work just fine, that is: Guests could connect - to host, - to internet - to other virtual maschines in the same subnet - all vice-versa. Internet can connect to any guest and our host can connect to any guest, too. Then, we needed one guest to connect to a guest from the other subnet. And this does not work, what I get follows: PING xx.xx.240.122 (xx.xx.240.122) 56(84) bytes of data. >From xx.xx.239.201 icmp_seq=1 Destination Port Unreachable >From xx.xx.239.201 icmp_seq=2 Destination Port Unreachable Note, that I can ping the other subnets gateway just fine! But I cannot connect to any of the other subnet IPs. I believe this is just a routing problem, but I cannot figure out what route to add where. Please ask if you need any more information.

2 1

[libvirt-users] (no subject)
by Abhishek Gupta 05 Jul '11

05 Jul '11

Hello I am having a libvirt installation on kvm, ubuntu 10.04. A few months ago, I setup a cloud on the hypervisors which was functioning really cool. But from last two-three days I am getting following error on using function snapshotListNames in python. >>> snapshots=dom.snapshotListNames(0) libvir: Remote error : unknown procedure: 187 Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/usr/lib/python2.7/dist-packages/libvirt.py", line 890, in snapshotListNames if ret is None: raise libvirtError ('virDomainSnapshotListNames() failed', dom=self) libvirt.libvirtError: unknown procedure: 187 Any ideas so as what could be causing the error? -- Abhishek Webpage <http://abhishekgupta92.info/> !!! Blog<http://thelazy.info/abhishek>

1 0

[libvirt-users] Associating a statefile with an image
by sethuraman subbiah 05 Jul '11

05 Jul '11

Hi , I am trying to perform offline migration (i.e) Create an incremental image using the qcow format, transfer the vm memory state to a state fie.Use the image and statefile together as a template. Now create a new vm using the template. I can successfully do it using the following commands : Save phase : stop migrate "exec:gzip -c > STATEFILE.gz" qemu-img qemu-img create -b BASE_img -f qcow2 INCRE_img Restore phase : qemu-kvm -m 1024 -hda INCRE_img -incoming "exec: gzip -c -d STATEFILE.gz" And it works fine. But I am not able to find the vm with virt-manager or other libvirt based tools. If I use : virsh save <dom_id> STATEFILE I can restore using "virsh restore STATEFILE" but I want to associate this with the incremental image that I created, but I dont know how. I think it saves the existing vm's xml file along with the STATEFILE, thus I am not able to change the disk image to the incremental image. Thus , 1. Either I should find a way to make vms created using qemu-kvm appear in libvirt-based tools. OR 2. Find a way to associate the virsh save STATEFILE with a incremental image. Any help or hint with respect to these will be very helpful. I am try to do cloning with minimal cost. This takes only 25 seconds ( to create STATEFILE) and creating incremental image is instantatenous . But I want to use the vm monitoring code based on libvirt. And it makes life difficult. Thanks for your help. - Regards, Sethuraman Subbiah Graduate Student - NC state University M.S in Computer Science

2 3