Scripts and experiment files for the puzzles experiment on the DETER testbed. These scripts are to be run on a swapped-in experiment on the testbed. For information about how to swap in/out experiments see the DETER documentation
For the rest of this document, we assume that the name of the project running on DETER is ILLpuzzle.
The script file run_experiment.sh provides a nifty way to run experiments and package their results in a single tar file.
The documentation for this script is show below
[Usage]: ./run_experiment.sh event_file [archive_name] [experiment_name]
Launch an experiment scenario and collect its results
@event_file: The AAL file containing the scenario definitions
@archive_name (Optional): The suffix to append to the name of the produced tarball. Default is a timestamp of the current day and time.
@experiment_name (Optional): The name of the DETER experiment that is swapped-in and to be used. Default is "oak"
For example, using
./run_experiment.sh exp1.yaml -exp1 myexp
will launch the scenario defined in exp1.yaml using DETER's magi orchestrator.
It will collect the pcap files generated in the directory /proj/ILLpuzzle/results and produce a tarball named results-exp1.tar.gz.
At this moment, this script only looks for pcap files, though it is easy to supplement it with the ability to tar other results files. Additionally,
the file will provide you with the option to cleanup the results directory before launching the scenario to avoid packaging differenet scenario files together.
DETER uses the magi orchestrator to deploy and launch the scenarios on the experiment machines. magi comes with a set of modules that allow you to run
different applications managed by the orchestrator, such as an Apache web server and clients, tcpdump, etc.
runcmd is a magi module that allows you to run a shell command as part of your scenario. We had to customize the version of the module provided by DETER
since it did not allow us to provide our the command to run with more than a single parameter. The custom version of the runcmd module can be found in
/proj/ILLpuzzle/modules/runcmd and is the one we use in all of our experiments.
In all of our experiments, we used a custom kernel patched to support client puzzles. The debian files needed to install the kernel onto an Ubuntu machine
can be found in /proj/ILLpuzzle/linux-4.13.0/. We also provide a script that installs our custom kernel on all of the machines in the experiment. To run
this script, simply use
bash push_kernel.sh
Note however that you might need to manually alter the script to include your custom experiment configuration.
Additionally, in order to ease the use of the magi orchestrator and its daemons on all of the experiment machines, we have added magi as a service on
all of the experiment machines. The script push_kernel.sh takes care of installing the service on each machine and configuring it to run when the system
boots. In case of any errors in the orchestration, you can simply manipulate the magi service on any machine using
sudo service magi [start|restart|stop|status]
Note however that restarting/stopping the magi service will kill all of its children. This is a feature of running the magi daemon as a service since
no orphaned processes will be left out in case of a failure (we had run into this issue multiple times with using the stock daemons).
Finally, the script push_remote_command.sh allows you to run a shell command on each of the experiment machines. The documentation for this script is as
follows
[Usage]: ./push_remote_command.sh cmd exp
@cmd: The command to run on all the machines, enclose with quotes in case it needs parameters
@exp: The name of the current experiment you are using