2568af3b98
Closes https://github.com/status-im/nim-waku/issues/565 |
||
---|---|---|
.. | ||
node | ||
protocol | ||
utils | ||
README.md |
README.md
Waku v2
This folder contains code related to Waku v1, both as a node and as a protocol.
Introduction
This is an implementation in Nim of Waku v2, which is currently in draft/beta stage.
See spec.
How to Build & Run
Prerequisites
- GNU Make, Bash and the usual POSIX utilities. Git 2.9.4 or newer.
- PCRE
More information on the installation of these can be found here.
Wakunode
# The first `make` invocation will update all Git submodules.
# You'll run `make update` after each `git pull`, in the future, to keep those submodules up to date.
make wakunode2
# See available command line options
./build/wakunode2 --help
# Connect the client directly with the Status test fleet
# TODO NYI
#./build/wakunode2 --log-level:debug --discovery:off --fleet:test --log-metrics
Waku v2 Protocol Test Suite
# Run all the Waku v2 tests
make test2
You can also run a specific test (and alter compile options as you want):
# Get a shell with the right environment variables set
./env.sh bash
# Run a specific test
nim c -r ./tests/v2/test_waku_filter.nim
Waku v2 Protocol Example
There is a more basic example, more limited in features and configuration than
the wakunode1
, located in examples/v2/basic2.nim
.
There is also a more full featured example in examples/v2/chat2.nim
.
Waku Quick Simulation
NOTE: This section might be slightly out of date as it was written for Waku v1.
One can set up several nodes, get them connected and then instruct them via the JSON-RPC interface. This can be done via e.g. web3.js, nim-web3 (needs to be updated) or simply curl your way out.
The JSON-RPC interface is currently the same as the one of Whisper. The only difference is the addition of broadcasting the topics interest when a filter with a certain set of topics is subcribed.
The quick simulation uses this approach, start_network
launches a set of
wakunode
s, and quicksim
instructs the nodes through RPC calls.
Example of how to build and run:
# Build wakunode + quicksim with metrics enabled
make NIMFLAGS="-d:insecure" wakusim2
# Start the simulation nodes, this currently requires multitail to be installed
# TODO Partial support for Waku v2
./build/start_network2 --topology:FullMesh --amount:6 --test-node-peers:2
# In another shell run
./build/quicksim2
The start_network2
tool will also provide a prometheus.yml
with targets
set to all simulation nodes that are started. This way you can easily start
prometheus with this config, e.g.:
cd ./metrics/prometheus
prometheus --config.file=prometheus.yml
A Grafana dashboard containing the example dashboard for each simulation node
is also generated and can be imported in case you have Grafana running.
This dashboard can be found at ./metrics/waku-sim-all-nodes-grafana-dashboard.json
To read more details about metrics, see next section.
Using Metrics
Metrics are available for valid envelopes and dropped envelopes.
To compile in an HTTP endpoint for accessing the metrics we need to provide the
insecure
flag:
make NIMFLAGS="-d:insecure" wakunode2
./build/wakunode2 --metrics-server
Ensure your Prometheus config prometheus.yml
contains the targets you care about, e.g.:
scrape_configs:
- job_name: "waku"
static_configs:
- targets: ['localhost:8008', 'localhost:8009', 'localhost:8010']
For visualisation, similar steps can be used as is written down for Nimbus here.
There is a similar example dashboard that includes visualisation of the
envelopes available at metrics/waku-grafana-dashboard.json
.
Spec support
This section last updated November 16, 2020
All Waku v2 specs, except for bridge, are currently in draft.
Docker Image
By default, the target will be a docker image with wakunode
, which is the Waku v1 node.
You can change this to wakunode2
, the Waku v2 node like this:
make docker-image MAKE_TARGET=wakunode2
docker run --rm -it statusteam/nim-waku:latest --help