• Install GNS3 by following the install instructions appropriate for your system
• Plug in router images in the GNS3 folder of your installation, on Linux, by default this is the directory ~/GNS3/images
• Run docker_setup.sh to generate the necessary docker images.
• Run the GNS3 server, by default this can be done by simply running gns3server in a terminal
• Then run run_batfish.sh if you want to run tests using Batfish. This uses the batfish docker image, it can later be resumed by using docker start batfish.
• Finally, run run_docker.sh to run the Metha docker image. By default, this will mount this directory into ~/metha inside the docker image for easy access to files there. This docker container can be resumed by running docker start metha-system after the initial generation.

• Install GNS3 by following the install instructions appropriate for your system
• Plug in router images in the GNS3 folder of your installation
• There are some scripts available for the installation of the following programs in the scripts folder. However, all prerequisites need to be installed first.
• Install PICT
• Install Batfish
• Install NV
• Install the NV fork of Batfish
• Install C-BPG
• Finally clone ntc-templates, and, if you did not clone it to ~, set the following environment variable:

export NET_TEXTFSM=/path/to/ntc-templates/templates/

• Clone this git repository and run

pip install -r requirements.txt

• Finally adjust config.json to point to your manually installed files

• Systems to be tested needs to be properly installed
• To test Batfish, additionally the Batfish server needs to be up and running. This can be done by running scripts/run_batfish.sh.
• The GNS3 Server needs to be properly installed and running
  • The router image used should be installed to the GNS3 image directory. By default, Metha supports the Cisco c7200-adventerprisek9-mz.124-24.T8.image image and Juniper VMX which requires junos-vmx-x86-64-18.2R1.9.qcow2, vmxhdd.img, metadata-usb-re.img, and vFPC-20180605.img to be installed in GNS3, i.e. to be placed in the images directory of GNS3.
  • Password protection of the GNS3 server needs to be disabled

To generate new test cases, run

python3 metha.py -p path run -s system

where path is the directory where the new test cases will be saved to after generation, and system is the system which is being tested, by default the supported systems are batfish, nv, cbgp. The GNS3 server needs to be running, by default Metha expects a local GNS3 server. In addition, to test Batfish the Batfish server also needs to be running locally. This will generate a results directory on the same level as path. In this directory, a summary of results will be saved as well as reports for any discrepancies found. In the folder passed by path, Metha will generate a subdirectory for every test case which includes all configurations as well as the computed routing tables in csv format.

Both the Batfish and GNS3 Servers must be running locally, then run

python3 metha.py -p path single-test -s system

where path is a directory containing at least a folder config of configs used for the routers, plus possibly a topology file (can be generated as well). If C-BGP is used, additionally there must also be a file called cbgp_config.txt in the folder at path.

More detailed documentation can be found in the documentation folder in this git repository. Specifically, the following documentation files are available:

• Command-Line-Options.md: Describes all different command-line options to running Metha, including how to run different evaluation configurations
• Configuration.md: Describes the Metha configuration in the config.json file. This sets up a lot of base configurations such as the location of external executables.
• Extensions.md: Describes how to extend Metha, in particular to different routers and to testing different systems.
• Network-Configuration-Features.md: Describes how to restrict the set of configuration features used by Metha.
• Topology.md: Describes how to define new topologies to run Metha on.

Metha is a project at the ICE Center at ETH Zurich, comprising researchers from the Networked Systems Group and the Secure, Reliable, and Intelligent Systems Lab:

• Rüdiger Birkner
• Tobias Brodmann
• Petar Tsankov
• Laurent Vanbever
• Martin Vechev