Unit-e Simulations

Tools to simulate Unit-e networks

License

This code published in this repository is released under the terms of the MIT license. See LICENSE for more information or see https://opensource.org/licenses/MIT.

Requirements

Python 3.6
Pipenv ( https://pipenv.readthedocs.io/en/latest/ )
- You can configure pipenv to create per-project virtual environments in the working directory instead of putting them in a shared directory by setting the environment variable PIPENV_VENV_IN_PROJECT=1.
A copy of the unit-e repository must be locally accessible (and the unit-e must be built)

Setup / Configuration

Copy the .env.example file to the .env file and ensure that the specified paths are correct and reachable (it will depend on your local environment, that's why .env is not versioned).
In order to develop the simulations, you will need to install the dev packages with the pipenv tool by executing pipenv install --dev.

Executing commands

Execute the command pipenv shell to enter into the virtual environment.
Execute pytest to run the tests & the static checks.
The experiments are in the experiments package, to run one of them just type ./experiments/experiment_name.py.

Running simulations & experiments

Using the C++ binaries

Here you can find a very basic example that allows us to create a local network using the C++-implemented nodes.

from asyncio import get_event_loop

from experiments.forking_simulation import ForkingSimulation

import test_framework.util as tf_util


# Because we use part of Unit-e's functional tests framework (specifically the
# TestNode wrapper), we have to set some global properties.
tf_util.MAX_NODES = 500  # has to be greater than 2n+2 where n = num_nodes
tf_util.PortSeed.n = 314159  # We want reproducible pseudo-random numbers


# In order to run a simulation, we'll create a `ForkingSimulation` instance that
# will manage everything for us.
simulation = ForkingSimulation(
    loop=get_event_loop(),
    latency=0.5,
    num_proposer_nodes=45,
    num_relay_nodes=5,
    simulation_time=600,  # Measured in seconds
    sample_time=1,  # Measured in seconds
    graph_model='preferential_attachment',
    block_time_seconds=16,  # Measured in seconds
    block_stake_timestamp_interval_seconds=4,  # Measured in seconds
    network_stats_file_name='network_stats.csv',
    nodes_stats_directory='/home/user/experiment_results/'
)

# Don't close the loop if you want to run more simulations after this one
simulation.safe_run(close_loop=False)

# Once the experiment is executed, the data has to be gathered from the
# `network_stats.csv` file and all the stats files generated individually by
# each node. Usually this is done using Pandas or other similar libraries.

Using the simplified Python code

Here you can find a very basic example that allows us to run a PoSv3 simulation using pure Python code (simulating several hours or days in much less time).

from blockchain.simnet import SimNet

simulated_network = SimNet(
    simulation_time=600,
    num_proposer_nodes=45,
    num_relay_nodes=5,
    num_coins_per_proposer=3,
    coins_amount=1000,

    time_between_blocks=16,
    block_time_mask=4,
    difficulty_adjustment_window=2048,
    difficulty_adjustment_period=1,

    max_future_block_time_seconds=600,
    latency=0.1,
    processing_time=0.001,

    # This parameter exists to check a vulnerability that depends on how we
    # pick the consensus parameters
    num_greedy_proposers=0,

    # In Bitcoin, the "past median timestamp" is used to validate the current's
    # block timestamp, not just the past block's timestamp (this is done to
    # avoid issues due to out-of-sync clocks). This parameter controls how to
    # compute such median timestamp.
    num_blocks_for_median_timestamp=13,
)

# Once this function call finishes, we'll be able to inspect its state to gather
# all the data we need.
simulated_network.run()

for node in simulated_network.nodes:
    pass # do something here

Troubleshooting

Import Error while running Jupyter notebooks

It turns out that Jupyter starts Python kernels at the path where the *.ipynb files are located. So, if you are using relative paths in your .env file, this is the most probable cause. It can be fixed just by using absolute paths instead.

Unable to install dependencies with pipenv

Could be that pipenv was installed using Python 2 instead of Python 3. In systems like Ubuntu, pip3 should be used instead of pip to install pipenv.

There are some tools that could help to install & manage Python utilities, like pipx.

`TypeError: init() got an unexpected keyword argument 'dirname'`

Notice: This is a temporary workaround.

The current ForkingSimulation is compatible with Unit-e v0.1, but not with the current Unit-e's master branch due to interface changes in the TestNode class.

Be sure to checkout the tag v0.1.0 to build the Unit-e binaries.

Unable to open socket

Sometimes, when the experiments spawn a considerable amount of nodes (and connections between them), we can suffer crashes due to the impossibility of opening new connections sockets.

This can be solved by increasing our open files limits. If you type ulimit -a in the console, you'll see your current limits, the output will be similar to this:

core file size          (blocks, -c) 0
data seg size           (kbytes, -d) unlimited
scheduling priority             (-e) 0
file size               (blocks, -f) unlimited
pending signals                 (-i) 126908
max locked memory       (kbytes, -l) 65536
max memory size         (kbytes, -m) unlimited
open files                      (-n) 1024
pipe size            (512 bytes, -p) 8
POSIX message queues     (bytes, -q) 819200
real-time priority              (-r) 0
stack size              (kbytes, -s) 8192
cpu time               (seconds, -t) unlimited
max user processes              (-u) 126908
virtual memory          (kbytes, -v) unlimited
file locks                      (-x) unlimited

The limit that is causing us problems is the one for "open files", so we can fix the problem by typing ulimit -n 8192.

License

dtr-org/unit-e-experiments

About

Resources

License

Code of conduct

Contributing

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Languages

Packages