Tracker: First Network Test

[alrevuelta]

Coming from these top level requirements this issue aims to break down in lower level requirements, all tasks required to accomplish the First Network Test presented in the issue. If you think any of the tasks have to be further discussed, feel free to open an issue.

Assumptions:

• Only relay protocol is supported
• Only nwaku client is tested
• All nodes are visible to each other.

Tracker: First Network Test

• Hardcoded configuration to spin up two hardcoded nwaku nodes:
  • Starting hardcoded kurtosis tests #1
• Configuration to spin up n configurable nwaku nodes.
• Connect nodes at startup, using the RPC API. Edit: Done in Prometheus+Graphana in Kurtosis and fixed node connection #9
• Include prometheus and grafana in the setup, with all config and a dashboard automatically provisioned. Edit: Done in Optimize interconnection, Prometheus+Graphana in Kurtosis and fixed node connection #10
  • Activate metrics in waku nodes and generate Prometheus targets #5
• A metrics such as node resource consumption - CPU, bandwidth and memory usage. Edit: Tackled with Added cadvisor to run.sh #76 , completed when integrated with Prometheus.
• Write a first experimental test, eg curl localhost:8003/metrics | grep waku_connected_peers and assert that after some time its >x. Edit: Done in Added tests in starlark #39
• Allow nodes to use a given nodekey so its peerId is known in advance before starting the node. Edit: Done in Configurable toml #6
• Incorporate discovery via discv5, conneting only small subset of nodes with staticnode(s) or discv5-bootstrap-node(s) and letting the peers discover each other. Done in Added node configuration in gennet.py #99
• Publish (from a random subset of nodes) messages of size s at a rate r. Done in Add WSL to Kurtosis as a container #19
• Measure reliability, using a seq counter to the message payload.
• Measure latency using the sender timestamp field.

cc @AlbertoSoutullo @Daimakaimura @LNSD
Let me know what you think.

[fryorcraken]

@kaiserd I believe this is mostly complete, right? Shall we ref to the 1k nodes final report here?