From CSLabsWiki
Revision as of 15:14, 17 August 2015 by Northug (talk | contribs)
Contact Person: CoreMaintainers
Last Update: Summer 2015
Services: Various

Operating system: Debian 8.0 Jessie
NIC 1: Clarkson Network
MAC: ?
CPU: 2x Intel Xeon E5410 @ 2.33GHz

Phoenix is our secondary VM host. It rose from the ashes of europa and xen2 It is currently in development.

Phoenix has older drives than eevee, but several of them have never been used prior to being installed in Phoenix. Additionally, all 4 drives, which are 1 TB each, have been configured in RAID6 to allow for up to 2 failed drives and provide a total of 2 TB of storage. This is twice the amount of storage as eevee currently has and also twice the tolerance for failure.


Generally, we followed the instructions on the debian wiki.

Things to install

apt-get install htop vim sudo qemu-kvm libvirt-bin virtinst

Network Bridge Setup

In order for the VMs to be on our LAN, a bridge must be set up as follows:

In /etc/network/interfaces:

# The loopback network interface
auto lo
iface lo inet loopback

auto br0
iface br0 inet static
	bridge_ports all  # or just eth0, ...

After making modifications to this file, you can either restart networking (/etc/init.d/networking restart, systemctl restart networking, etc.) or use ifup br0

Manual set up:

1) Create new bridge iterface "brctl addbr <bridge_name>"

2) Add existing interface to bridge "brctl addif <bridge_name> <iterface_name>"

3) Bring bridge up "ifconfig <bridge_name> up"

Bridging interfaces will cause the interfaces used to not have an ip address even if they are up. The bridge interface will contain an ip address.

Migration from eevee

To migrate a VM from eevee (in the case that it was created by kimchi):

Put iso in /var/lib/libvirt/isos

Put xml file from dump into /var/lib/libvirt/qemu

(The dump can be had by running virsh dumpxml > vm_name.xml on eevee

XML Edits

You will, at the least, need to make sure the network interface is running in bridge mode. In <devices/>, make sure that the <interface/> node resembles something of the following:

    <interface type='bridge'>
      <mac address='00:16:36:15:a1:dd'/> <!-- Usually generated, can be omitted for a random one IIRC -->
      <source bridge='br0'/>
      <model type='virtio'/> <!-- Other models may work, but are untested -->
      <address type='pci' domain='0x0000' bus='0x00' slot='0x04' function='0x0'/> <!-- Probably also safe to omit -->

If you're doing this to an already running VM, you will need to completely shut it down and restart it--a reboot will not suffice. When in doubt, shutdown, destroy, start. The network adapter for the VM should now show up as one of the vnet# devices, and will be automatically assigned to br0 (as above).

Put disk image in /var/lib/libvirt/images

virsh define <Your VM Name>.xml