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.
- 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
- 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
- Arbitrary lightning daemon binary detection (lnd, clightning, ...)
- Time-dependent transaction series
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
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.
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
To run all tests, run python3 -m unittest discover test from the root folder.
$ docker build -t lnregtest:local .
$ docker run --rm -it lnregtest:local python3 -m unittest discover test
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.
- Medium article on how regtest lightning networks can be set up: bitstein-medium
- Dockerized lightning networks: simverse
- Dockerized version for the medium article: bitstein-test-env