Wednesday, January 17, 2018

Cumulus - Libvirt and KVM - QEMU


Source: https://docs.cumulusnetworks.com/display/VX/Libvirt+and+KVM+-+QEMU

The following sections describe how to set up a two-leaf/two-spine Cumulus VX topology with QEMU/KVM on a Linux server.

These sections assume a basic level of Linux and KVM experience. For detailed instructions, refer to the QEMU and KVM documentation. 

Contents

Overview

Performing virtualization in Linux requires three components:
  • Libvirt provides an abstraction language to define a VM. It uses XML to represent and define the VM.
  • KVM works exclusively with QEMU and performs hardware acceleration for x86 VMs with Intel and AMD CPUs. The pair is often called KVM/QEMU or just KVM. 
  • QEMU is a machine emulator that allows the host machine to emulate the CPU architecture of the guest machine. Because QEMU does not provide hardware acceleration, it works well with KVM.

Install libvirt

  1. Review the Linux version of the host:

    This guide is validated and verified for Ubuntu Trusty 14.04.5 LTS starting from a clean install.
    local@host:~$ uname -a
    Linux ubuntu-trusty-64 3.13.0-93-generic #140-Ubuntu SMP Mon Jul 18 21:21:05 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux
  2. Run the following commands to install libvirt:
    local@host:~$ sudo add-apt-repository ppa:linuxsimba/libvirt-udp-tunnel
    local@host:~$ sudo apt-get update -y
    local@host:~$ sudo apt-get install libvirt-bin libvirt-dev qemu-utils qemu
    local@host:~$ sudo /etc/init.d/libvirt-bin restart

    The linuxsimba/libvirt-udp-tunnel package repository provides an updated libvirtd version that includes enhancements required to launch Cumulus VX. The example below shows the installation output:
    local@host:~/$ sudo apt-get install libvirt-bin libvirt-dev qemu-utils qemu
    Reading package lists... Done
    Building dependency tree      
    Reading state information... Done
    tree is already the newest version.
    git is already the newest version.
    qemu-utils is already the newest version.
    qemu-utils set to manually installed.
    The following packages were automatically installed and are no longer required:
      bsdtar libarchive13 liblzo2-2 libnettle4 linux-headers-4.2.0-34
      linux-headers-4.2.0-34-generic linux-image-4.2.0-34-generic
      linux-image-extra-4.2.0-34-generic ruby-childprocess ruby-erubis ruby-ffi
      ruby-i18n ruby-log4r ruby-net-scp ruby-net-ssh
    Use 'apt-get autoremove' to remove them.
    The following extra packages will be installed:
      libvirt0 python-paramiko python-support qemu-system qemu-system-arm
      qemu-system-mips qemu-system-misc qemu-system-ppc qemu-system-sparc
      qemu-user sshpass
    Suggested packages:
      radvd lvm2 qemu-user-static samba vde2 openbios-ppc openhackware qemu-slof
    The following NEW packages will be installed:
      htop python-paramiko python-support qemu qemu-system qemu-system-arm
      qemu-system-mips qemu-system-misc qemu-system-ppc qemu-system-sparc
      qemu-user sshpass
    The following packages will be upgraded:
      ansible libvirt-bin libvirt-dev libvirt0
    4 upgraded, 12 newly installed, 0 to remove and 18 not upgraded.
    Need to get 31.1 MB of archives.
    After this operation, 166 MB of additional disk space will be used.
    Do you want to continue? [Y/n] Y
  3. After the installation process is complete, log out, then log back in to verify the libvirt version.

    In this guide, libvirt 1.2.16 was verified.
    local@host:~# libvirtd --version
    libvirtd (libvirt) 1.2.16

Configure Cumulus VX VMs with QEMU/KVM


This section assumes that you have installed QEMU/KVM and the Cumulus VX disk image for KVM. For download locations and steps, refer to the Getting Started page.

This configuration is tested on a server running Debian 3.2.60-1+deb7u3 x86_64 GNU/Linux with 3.2.0-4-amd64 #1 SMP processors.
After you follow the steps below, the interfaces will be connected as follows:
  • leaf1:swp1--->spine1:swp1
  • leaf1:swp2--->spine2:swp1
  • leaf2:swp1--->spine1:swp2
  • leaf2:swp2--->spine2:swp2
  1. Copy the qcow2 image onto a Linux server four times to create the four VMs, then name them as follows:
    • leaf1.qcow2
    • leaf2.qcow2
    • spine1.qcow2
    • spine2.qcow2
  2. Power on leaf1.qcow2 and configure it as follows:
    sudo /usr/bin/kvm   -curses                             \
                        -name leaf1                       \
                        -pidfile leaf1.pid                \
                        -smp 1                              \
                        -m 256                              \
                        -net nic,vlan=10,macaddr=00:01:00:00:01:00,model=virtio \
                        -net user,vlan=10,net=192.168.0.0/24,hostfwd=tcp::1401-:22 \
                        -netdev socket,udp=127.0.0.1:1602,localaddr=127.0.0.1:1601,id=dev0 \
                        -device virtio-net-pci,mac=00:02:00:00:00:01,addr=6.0,multifunction=on,netdev=dev0,id=swp1 \
                        -netdev socket,udp=127.0.0.1:1606,localaddr=127.0.0.1:1605,id=dev1 \
                        -device virtio-net-pci,mac=00:02:00:00:00:02,addr=6.1,multifunction=off,netdev=dev1,id=swp2 \
                        leaf1.qcow2
  3. Power on leaf2.qcow2 and configure it as follows:
    sudo /usr/bin/kvm   -curses                             \
                        -name leaf2                       \
                        -pidfile leaf2.pid                \
                        -smp 1                              \
                        -m 256                              \
                        -net nic,vlan=10,macaddr=00:01:00:00:02:00,model=virtio \
                        -net user,vlan=10,net=192.168.0.0/24,hostfwd=tcp::1402-:22 \
                        -netdev socket,udp=127.0.0.1:1604,localaddr=127.0.0.1:1603,id=dev0 \
                        -device virtio-net-pci,mac=00:02:00:00:00:03,addr=6.0,multifunction=on,netdev=dev0,id=swp1 \
                        -netdev socket,udp=127.0.0.1:1608,localaddr=127.0.0.1:1607,id=dev1 \
                        -device virtio-net-pci,mac=00:02:00:00:00:04,addr=6.1,multifunction=off,netdev=dev1,id=swp2 \
                        leaf2.qcow2
  4. Power on spine1.qcow2 and configure it as follows:
    sudo /usr/bin/kvm   -curses                             \
                    -name spine1                       \
                    -pidfile spine1.pid                \
                    -smp 1                              \
                    -m 256                              \
                    -net nic,vlan=10,macaddr=00:01:00:00:03:00,model=virtio \
                    -net user,vlan=10,net=192.168.0.0/24,hostfwd=tcp::1403-:22 \
                    -netdev socket,udp=127.0.0.1:1601,localaddr=127.0.0.1:1602,id=dev0 \
                    -device virtio-net-pci,mac=00:02:00:00:00:05,addr=6.0,multifunction=on,netdev=dev0,id=swp1 \
                    -netdev socket,udp=127.0.0.1:1603,localaddr=127.0.0.1:1604,id=dev1 \
                    -device virtio-net-pci,mac=00:02:00:00:00:06,addr=6.1,multifunction=off,netdev=dev1,id=swp2 \
                    spine1.qcow2
  5. Power on spine2 and configure it as follows:
    sudo /usr/bin/kvm   -curses                             \
                    -name spine2                       \
                    -pidfile spine2.pid                \
                    -smp 1                              \
                    -m 256                              \
                    -net nic,vlan=10,macaddr=00:01:00:00:04:00,model=virtio \
                    -net user,vlan=10,net=192.168.0.0/24,hostfwd=tcp::1404-:22 \
                    -netdev socket,udp=127.0.0.1:1605,localaddr=127.0.0.1:1606,id=dev0 \
                    -device virtio-net-pci,mac=00:02:00:00:00:07,addr=6.0,multifunction=on,netdev=dev0,id=swp1 \
                    -netdev socket,udp=127.0.0.1:1607,localaddr=127.0.0.1:1608,id=dev1 \
                    -device virtio-net-pci,mac=00:02:00:00:00:08,addr=6.1,multifunction=off,netdev=dev1,id=swp2 \
                    spine2.qcow2

    The QEMU/KVM commands used here are minimal. You can add more parameters, such as -enable-kvm, -serial or -monitor, as needed.
    Bridging Switch Port Interfaces

    If you intend to bridge the switch ports in the VM, place each switch port in the bridge in its own virtual network on the host. Otherwise, you might see this error:
    br0: received package on swp1 with own address as source address

Next Steps


This section assumes that you are configuring a two-leaf/two-spine network topology, that you have completed the steps in Create a Cumulus VX Virtual Machine with VMware vSphere - ESXi 5.5 above, and that you now have a VM called CumulusVX-leaf1.
After you create all four VMs, follow the steps in Create a Two-Leaf, Two-Spine Topology to configure the network interfaces and routing.