Skip to content

Regtest Lightning Networks for (python) integration testing

Notifications You must be signed in to change notification settings

bolt-observer/lnregtest

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

70 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

lnregtest - Lightning Networks on Bitcoin regtest

Implements functioning lightning networks for (Python) integration testing operating on a Bitcoin regtest network by running several lightning node instances.

The simulated lightning networks can have different shapes as defined in the network_definitions folder.

An example of a graph (star_ring) with 7 nodes and 12 channels is shown here (for initial channel balance details, have a look at network_definitions/star_ring.py):

Star-like component with channels (where A is the master node):
A -> B, A -> C, A -> D, A -> E, A -> F, A -> G,
Ring-like component with channels (which surrounds the master node):
B -> C, C -> D, D -> E, E -> F, F -> G, G -> B

This star and ring-like lightning network can then be used to test interactions with the network from the master node's perspective, like rebalancing channels, routing payments, sending payments and so on.

Features

  • No external python dependencies
  • Arbitrary lightning network graphs with up to number of nodes on the order of 10 (depends on your resources)
  • LND support
  • Core lightning support
  • Electrum support
  • Lightning graph state comparison
  • Restarting from already created networks
  • Abstraction of random identifiers (public keys, channel ids) to human readable identifiers
  • Library and command-line execution
  • Automatic sanity check of user defined networks

Planned features

  • Arbitrary lightning daemon binary detection (lnd, clightning, ...)
  • Time-dependent transaction series

Creating your own network topology

Networks of arbitrary shape can be defined as a python dictionary in the network_definitions folder. See the examples for a general structure.

The requirements are:

  • Node A is the master node
  • Nodes are uniquely named by subsequent characters A, B, C, ...
  • Channels are uniquely numbered by integers 1, 2, 3, ...
  • Ports must be set uniquely

Testing other (python) projects

Test examples can be found in the test folder and more information on how lnregtest is used for lightning network integration testing can be found in TEST.

This package is also used in production in lndmanage.

Setup

The binaries bitcoind (v22.0), bitcoin-cli, lnd (v0.14), and lncli are expected to be found in $PATH, e.g., put these binaries into your ~/bin folder.

You can use the tool in two different standalone modes:

Git repository mode:

$ git clone https://github.com/bitromortac/lnregtest
$ cd lnregtest

Run network:

$ python3 lnregtest.py -h

Package mode:

$ python3 -m venv venv
$ source venv/bin/activate
$ python3 -m pip install lnregtest

Run network:

$ lnregtest -h

Test if lnregest works

To run all tests, run python3 -m unittest discover test from the root folder.

Running under Docker

$ docker build -t lnregtest:local .
$ docker run --rm -it lnregtest:local python3 -m unittest discover test

NOTE for ARM-based system users

If you are using ARM-based system such as MacBook M1s and M2s, make sure to select amd64 for your build command as followed

$ docker build -t lnregtest:local --platform linux/amd64 .
...

Troubleshooting

  • all SubConns are in TransientFailure: Typically, here it happens that lnd is not given enough time to start. The simulation of a lightning network is memory and CPU intensive. Each LN node needs some time to get up and running and consumes resources. Currently, the startup of each lnd node is delayed to distribute CPU load. The settings were tested on a quadcore processor and 8 GB of RAM.
  • bitcoind and lnd processes are not terminated: Sometimes it happens that the processes are not terminated correctly, so before you start new tests, make sure to do so manually.

Related Projects

About

Regtest Lightning Networks for (python) integration testing

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 95.4%
  • Dockerfile 3.8%
  • Shell 0.8%