- Users - libvirt List Archives

Re: [libvirt-users] About live migration with snapshots
by Edward Young 11 Feb '15

11 Feb '15

Hi All, Sorry, I do not know to reply the previous message directly, so I have to manually copy them here. I have two questions about the following issue? 1. I follow the instructions blew to migrate a vm with snapshots. When I perform 'virsh snapshot-create --redefine $dom file' on the destination.' I got an error saying that "no domain with matching name $dom". I'm wondering during the mgiration, what do we need to do in the destination? create a new base file? 2. What does the following instruction do? "you then loop over 'virsh snapshot-delete --metadata $dom $name' on the source, at which point, live migration will now work." 3. Also, could you explain a little about the high level techniques for this function. suppose we have a vm with one base file and two snapshots. How does the following instructions do to migrate all of them? Thanks! Edward ------------------------------ - *From*: Eric Blake <eblake redhat com> - *To*: Chiang Hubert <clhtwn gmail com> - *Cc*: libvirt-users redhat com - *Subject*: Re: [libvirt-users] About live migration with snapshots - *Date*: Wed, 20 Mar 2013 21:40:26 -0600 ------------------------------ On 03/20/2013 08:45 PM, Chiang Hubert wrote: > Hello, > > I'd like to live migration with snapshots. > > But it doesn't work. It comes out a message "cannot migrate domain with 1 > snapshots" Unfortunately, figuring out how to migrate snapshot information at the libvirt API level requires some engineering work - the current RPC protocol for migration is not set up to migrate an arbitrary amount of snapshots in a single call. On the other hand, if you are allowed to make more than one API call, the solution is already available; maybe we should patch virsh to learn how to make the series of API calls, to automate what I will describe below. > > Then I try to trace the code(Libvirt 0.9.8 to 1.0.3), I find out the code > in src/qemu/qemu_migration.c @ Line 1395 - 1440 (Libvirt 1.0.3) It's still unimplemented at the libvirt level, even in libvirt.git. > > It will check the VM which has snapshots or not. > > I just curious about this limitation, why the VM can't live migration with > snapshots? Doing it all in one RPC call would be a potential denial-of-service (RPCs are bounded in length to avoid consuming server resources, and taking lots of snapshots on the source could easily be made to exceed bounds). If someone can design a way to set up a series of RPC handshakes, then we could do it at the libvirt level in a single API call, but I'm not sure it is worth it. > > What happen if I skip this check? > > Does it has any suggestion way or virsh command with options to do live > migration with snapshots? The existing solution at the management tool layer is to migrate the snapshot information first, and then to migrate the domain. For each snapshot in 'virsh snapshot-list --name $dom', you will want to 'virsh snapshot-dumpxml $dom $name > file' on the source, then 'virsh snapshot-create --redefine $dom file' on the destination. Next, determine 'virsh snapshot-current --name $dom' on the source, and use 'virsh snapshot-current $dom $name' on the destination to set it as current (if there is a current snapshot). After the destination has all the snapshots, you then loop over 'virsh snapshot-delete --metadata $dom $name' on the source, at which point, live migration will now work. Patches to teach virsh how to do all this work in a single 'virsh migrate' are welcome. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org *Attachment: signature.asc <https://www.redhat.com/archives/libvirt-users/2013-March/pgpmxFegVXm2O.pgp>* *Description:* OpenPGP digital signature

1 0

[libvirt-users] Possible to change 'machine type' attribute while vm is running?
by Andre Goree 09 Feb '15

09 Feb '15

Hello. I'd like to determine if it's possible to change the 'machine type' attribute while a vm is running. For example, I have: <os> <type arch='x86_64' machine='pc-i440fx-trusty'>hvm</type> <boot dev='hd'/> </os> I'd like to change that to "machine='pc-i440fx-1.7'" due to an error[1] I'm seeing while migrated similar images to another host. The only way I've been able to change the machine is by first shutting down the vm, making the change, then starting the vm back up. I'd like to avoid that if at all possible. I did find the tool "libvirt-migrate-qemu-machinetype", but it doesn't seem to do what I think it's intended to do -- i.e., it does NOT change the machine type, even after reboot. Perhaps I'm just using it wrong? I performed the following: root:~# libvirt-migrate-qemu-machinetype -t pc-i440fx-1.7 <domain> Waiting up to 10 seconds for libvirtd to start... Waiting up to 30 seconds for libvirtd to respond to requests... Checking domains defined in /etc/libvirt/qemu... Migration complete But the machine type remains the same after the above command has been run. Thanks in advance for any insight you can provide. [1] - error: internal error: process exited while connecting to monitor: Supported machines are: -- Andre Goree -=-=-=-=-=- Email - andre at drenet.net Website - http://www.drenet.net PGP key - http://www.drenet.net/pubkey.txt -=-=-=-=-=-

2 1

[libvirt-users] CPU model and type
by Paul Apostolescu 06 Feb '15

06 Feb '15

Hello, Is there a way to configure the domain cpu in such a way that the info reported to the guest OS system will remain constant ? For example in older versions of of libvirt/qemu the cpu was reported as "QEMU Virtual CPU version (cpu64_rhel6)" but moving the vm on a qemu2.2.0 is is reported as "QEMU Virtual CPU version 2.2.0". Thanks.

2 2

[libvirt-users] CPU model and missing AES-NI extension
by Dennis Jacobfeuerborn 06 Feb '15

06 Feb '15

Hi, today I tried to configure a guest using Virt-Manager and used the "copy host cpu configuration" option which resultet in a "Sandy Bridge" model. What I noticed is that for example the "aes" extension is not available in the guest even though it is available on the host cpu. This is what the host cpu looks like: model name : Intel(R) Xeon(R) CPU E5-2650 v3 @ 2.30GHz flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid And this is what I get in the guest: model name : Intel Xeon E312xx (Sandy Bridge) flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss syscall nx pdpe1gb lm constant_tsc pni ssse3 cx16 sse4_1 sse4_2 popcnt lahf_lm abm Is there a way to get closer to the hardware and get as many of the host features in the guest as possible? I never intend to do anything like live-migration with this guest so compatibility for that case is not a concern. Regards, Dennis

2 2

Re: [libvirt-users] vnc client for libvirt from os x?
by shaowt 05 Feb '15

05 Feb '15

1 0

[libvirt-users] vnc client for libvirt from os x?
by pixelfairy 05 Feb '15

05 Feb '15

id like to access libvirt kvm consoles from my macbook. i tried realvnc, which just crashes immediately, and the built in "screen sharing" client, which insists on a password, which i havent set. does anyone here do this? which vnc client do you use?

2 3

[libvirt-users] Vepa use vf?
by Yaron Illouz 02 Feb '15

02 Feb '15

1. Does vepa mode in libvirt use sr-iov ? 2. How can I do port mirroring with sr-iov? vepa This network uses a macvtap "direct" connection in "vepa" mode to connect each guest to the network (this requires that the physical interfaces used be connected to a vepa-capable hardware switch. The physical interface to be used will be picked from among those listed in <interface>subelements of the <forward> element; multiple guest interfaces can share each physical interface (libvirt will attempt to balance usage between all available interfaces).Since 0.9.4

3 2

[libvirt-users] virsh snapshot-create persistent through destroy(undefine)
by Yitao Jiang 30 Jan '15

30 Jan '15

Hi, If i take a snapshot of a qcow2 image backed domain via virsh snapshot-create domainID and I shutdown the machine (and probably destroy the domain if virsh list still shows it running), and then undeinfed the domain will that snapshot still exist and be active when I define and start the domain back up? --- Thanks, Yitao(依涛姜) jiangyt.github.io

2 3

[libvirt-users] Disk space exhaustion
by Fabrizio Soppelsa 30 Jan '15

30 Jan '15

Hello, what happens when an instance writes data with 0% available physical disk space, on QCOW2 images for instance, apart from destruction and unpredictability? I'm aware that monitoring is mandatory in production environments, but are there other known industry tricks to deal with this? Soft/hard limit settings, autopause/stop, or does KVM react in doing force shut down the VM in question? Thank you, Fabrizio.

2 1

[libvirt-users] redistribute virtual machines from vps hosting to end users
by Vasiliy Tolstov 29 Jan '15

29 Jan '15

Hi. I need to get ability to download backups of kvm virtual machines (raw images) to end users. Virtualbox users can import ovs images, that i can create, but in case of linux and virt-manager - how users can import images and simply run it? I think about newbie linux users, that can install virt-manager under ubuntu via graphical package manager and do some easy steps... Can you share me info how can i do needed things ?=)Or may be some admins/users already solve this kind of "problems"... -- Vasiliy Tolstov, e-mail: v.tolstov(a)selfip.ru jabber: vase(a)selfip.ru

3 6