Tag Info

New answers tagged

1

The problem was using a disk image from DreamSpark/OnTheHub. I downloaded a new image, and it worked out of the box. The old image name was en_windows_server_2012_r2_x64_dvd_2707946.iso. I "ordered" a new free Windows Server 2012 of DreamSpark. I believe redownloading using Secure Download Manager will download the old image again, but I haven't tested ...


2

<disk type='file' device='disk'> <driver name='qemu' type='qcow2'/> <source file='/home/thomas/kvm/storage_pool/windows2012-test-3.qcow2'/> <target dev='vda' bus='virtio'/> </disk> This is what it should look like. BTW, yout NIC is also RTL and not virtio, tsk tsk tsk.


0

All you need to set up on the host is the bridge. When you configure virtual NICs, and attach them to the VM, the taps will get created automatically, when the VM is started. Now, without reading the actual guide, it looks like you need three interfaces on three DIFFERENT networks. If you plug all your virtual NICs into the same bridge (effectively, a ...


1

For your specific use case (base image + qcow2 overlay), the RAW format should be preferred: It's faster: as it has no metadata associated, it is as fast as possible. On the other hand, Qcow2 has two layer of indirection that must be crossed before to hit the actual data As the overlay layer must be a Qcow2 file, you don't lose the ever-useful snapshot ...


1

If the VM has lots of memory, saving it will in any case mean a large amount of time spent saving the memstate. If there is no hard requirement to backup a full VM state (because usually it is redundant, you'll get errors when you restore because of time differences, and it might even lead to a crash). Normally, VMs are backup up as following: Quiesce ...


0

Latest virt-manager (1.1.0) supports setting architecture on VM creation. You can not change it later. With virsh edit, I am guessing a little, but I would try removing the PCI controller and all bus related parameters (like <address />. And changing <type arch='i686' machine='pc-i440fx-2.1'>hvm</type> to <type arch='i686' ...


0

The error is misleading: it is in fact because qemu can't connect to your X session. To run X programs as a different user you need to either preserve your environment: sudo -E kvm ... or use xhost, e.g. xhost +si:localuser:root or set root's XAUTHORITY environment variable to point to your $HOME/.Xauthority file.


0

Stop the VM. Edit the domxml using virsh edit VMNAME and point the disk at the location of snap2 Start the VM. Don't forget to delete snap3 EDIT: branching out In order to be able to switch back to a snapshot in a chain, without corrupting the entire chain after the current snap, you need to create a branch. Using your example, you stop the VM, take a ...


0

I noticed that, unlike the other VM's, this VM's image was owned by root:root. I tried to chown it to libvirt-qemu:kvm like the other VM's, but the ownership changed back to root:root automatically. However, this did cause the VM to be in "shut off" state. Starting it with sudo virsh start vph-share brought it back to life.


1

This feature was requested a long time ago. Now libvirt supports it by providing two new commands: domifaddr and net-dhcp-leases Usage: domifaddr <domain> [interface] [--full] [--source lease|agent] Example outputs: virsh # domifaddr f20 --source agent Name MAC address Protocol Address ...


2

Qcow2 file backend can be significantly slow with cache=none setting. Moreover, "-o prellocation=metadata" preallocate metadata only, and actual file data will be fragmented. In other words, the qcow2 file remain a sparse one with only short stroke of allocation (for metadata). In the past a "-o preallocation=full" option appeared, by in recent qemu-img ...


2

cache=none is probably not a good idea when you're using qcow2 files. A qcow2 file makes it appear that every access to the disk is fragmented. This means you get the random access performance of the drive every time and some flash drives are awefully slow (spelling intended) at random writes. Try with cache=unsafe (temporarily) to confirm this is the ...


0

ovirt-engine, the controller node doesn't need KVM. ovirt hosts will install KVM and other packages, when added to the setup Just set up a minimal Linux server and add it via the GUI, everything will be bootstrapped OVS doesn't come with ovirt, ovirt uses Linux bridges by default.


0

You can create a small service to install on all VMs and configure them to execute that service at whatever moment you consider your boot process to be complete. That service can either hit an HTTP URL hosted by your benchmarking app (and telling its hostname and other stuff in the request data so the bench app knows which VM is which) or spit something out ...


0

The issue is that since this is a VM there is a partition from the image now transfered to the LVM. When mounting the LVM locally, you need to mount with an offset. You need to obtain the start of your partition within the disk. This will print out your partitions. Take the start # of your boot partition. parted -s /dev/vg00/oes2 unit s print example: ...


0

Yes, every snapshot metadata has the parent's filename or device name. But do keep in mind that the chain must be complete, e.g. snap2 is not aware of base, only of snap1


0

I've done extensive testing on KVM and caching performance (you can read here, here and here) and many of the recommendation that you find on the Internet is obsolete of plain wrong. But let's proceed one step at a time... RAID5 needs a much smaller chunk size (in the order of 32-64K) than yours (512K, as seen from pastebin) and a BBU (for writeback ...


0

You gave us quite no detail on how you use LVM in your setup. If your issue is on reads, their could be two origins. First of all, check if you're not using LVM snapshots, as they may (will on long term) drastically slow down IO as they grow. Secondly, your top states that your hypervisor is only using 1G of RAM for IO caches, which isn't really much, ...


1

Yes, all you need to do is take a snapshot of the disks while the VM is running and copy the backing image while the VM is running off the snapshot. To get rid of the snapshot live, you need to have a fresh version of libvirt and qemu, and you need to either execute virsh blockcommit or virsh blockpull, depending on whether you want the snapshot merged ...


-1

This is completely doable with RAW disk image format. I don't know about qcow2... Debian.img: x86 boot sector; partition 1: ID=0x83, active, starthead 32, startsector 2048, 497664 sectors; partition 2: ID=0x5, starthead 59, startsector 501758, 104353794 sectors, code offset 0x63 Debian2.img: x86 boot sector; partition 1: ID=0x83, active, ...


0

If you don't care about in-memory data, you can snapshot the backing device and take a copy of that snapshot. This of course require a storage pool with snapshot support (mostly LVM or ZFS, as BTRFS is abysmal slow for virtual machines storage).


0

Here is an example how you can test if a machine is online using ping: #!/bin/bash # test-online.sh ret=1 while [ $ret != 0 ] do ping -c 1 $1 2>/dev/null ret=$? done exit 0 You can adapt this to use ssh by replacing the ping line with something like ssh $1 "echo", assuming you can connect to the machine without a password. Using the above script ...


0

xml for hub: https://libvirt.org/formatdomain.html#elementsHub An example: <hub type='usb'> <address type='usb' bus='0' port='1'/> </hub> https://www.redhat.com/archives/libvir-list/2011-August/msg00816.html


0

Seems like you could use QEMU's -netdev socket option here, probably UDP version will fit better. Also -netdev vde may work, although is more complicated and requires VDE switch daemon configured. But I would try to overcome problem with Linux bridge anyway, with STP option enabled, as @Martin suggested. This is simple and very common setup.



Top 50 recent answers are included