Skip to content

VIRTUALIZING.md

Latest commit

 

History

History
131 lines (101 loc) · 4.81 KB

VIRTUALIZING.md

File metadata and controls

131 lines (101 loc) · 4.81 KB

Run LibreMesh in a virtual machine

In this repository, you will find scripts that allow you to raise libremesh instances on virtual machines using QEMU. This way you can test libremesh without the need to count with routers or to flash the libremesh firmware on them, which greatly speeds up the libremesh development process.

What are the limitations of virtualization?

This image is not a perfect firmware image, it does not have wifi for example but ethernet network LAN and WAN is supported. All the files inside the packages files/ can be copied into the rootfs, overwriting a precooked image that is a full LibreMesh x86_64 image. So don't expect that everything runs exactly as in a wireless router but most things will perform as expected:

  • initialization scripts: uci-defaults, init.d, etc
  • lime-config
  • ubus / rpcd
  • lime-app

ICMPv4 does NOT work between qemu nodes, so ping (v4) will not work as expected. Everything else (including ICMPv6 i.e. ping6) does work as expected, however.

What do you need to run libremesh in a virtual?

In order to run libremesh in QEMU you need two files:

  • generic-rootfs.tar.gz
  • generic-kernel.bin Which can be generated by compiling libremesh from the buildroot selecting x86 as the target and the option to generate ramgs.bzImage.

LibreRouterOS, a LibreMesh flavor maintained by the LibreRouter team, distributes these images in each of its releases. You can find them at: https://gitlab.com/librerouter/librerouteros/-/releases Example: https://repo.librerouter.org/lros/releases/1.5/targets/x86/64/

How to virtualize a libremesh node?

  1. Clone the lime-packages repository
git clone [email protected]:libremesh/lime-packages.git
  1. Install the following packages
sudo apt install dnsmasq iptables qemu-system-x86
  1. Run the script ./tools/qemu_dev_start
cd lime-packages
sudo ./tools/qemu_dev_start /path/to/rootfs.tar.gz /path/to/generic-kernel.bin
  1. To stop the virtual and clean the environment created on the host for its operation:
sudo ./tools/qemu_dev_stop

In addition, up to 100 qemu nodes can be configured. This can be done using the --node-id N parameter. In this example, the node id is 1 and the host network interface is wlo1:

sudo ./tools/qemu_dev_start --node-id 1 --enable-wan wlo1 /path/to/rootfs.tar.gz /path/to/generic-kernel.bin

All LAN interfaces of the node are tied together. You can use --enable-wan on just one of the nodes to share its internet connection with the network

How do I go about giving it access to the internet?

If you want to give the node access to the internet, qemu_dev_start argues “enable-wan” to which you must pass the name of the network interface of your host through which you have an internet connection as a parameter. In this example the wifi interface name is wlo1. The command is used as follows:

sudo ./tools/qemu-dev-start --enable-wan wlo1 path/to/rootfs.tar.gz /path/to/generic-kernel.bin

How do I stomp the image with my libremesh working directory?

If you've been working on libremesh mods and want to test them on a virtual node, you can use the "libremesh-workdir" argument, and step into the root file system without the need to rebuild the images. The command is used as shown next:

sudo ./tools/qemu-dev-start --libremesh-workdir . /path/to/rootfs.tar.gz /path/to/generic-kernel.bin

How do I virtualize a multi-node libremesh network?

  1. Install ansible
sudo apt install ansible

or by following the official ansible installation guide.

  1. Copy the rootfs.tar.gz and generic-kernel.bin images to ./tools/ansible/files
cd lime-packages
cp /path/to/rootfs.tar.gz ./tools/ansible/files/generic-rootfs.tar.gz
cp /path/to/generic-kernel.bin ./tools/ansible/files/generic-kernel.bin
  1. Run the playbook qemu_dev_start.yml
cd lime-packages/tools/ansible
sudo ansible-playbook qemu_cloud_start.yml
  1. To stop the virtual ones and clean the environment created on the host for its operation:
sudo ansible-playbook qemu_cloud_stop.yml

By default, the network topology is made up of 12 nodes distributed in four different batman/L2 clouds (A, B, C, D) and interconnected between them. To modify this topology, you can edit the ./tools/ansible/hosts.yml file.

You will then be able to access each of the cloud nodes via clusterssh as shown in the playbook output.

How do I specify in which node of the mesh the internet connection would be?

Edit the hosts.yml file and add the enable_wan variable with the value of the network interface of your host through which you have an internet connection. Example:

all:
  children:
    cloudA:
      vars:
        eth0: lm_cloudA
        cloud: A
      hosts:
        hostA.cloudA.test:
            enable_wan: wlo1
(...)