prkl project feedback
jrhea opened this issue · 2 comments
jrhea commented
Creating an issue to collect suggestions/feedback/priorities from others
jrhea commented
Maybe a reasonable goal for v0.2 testnet is to try and map out the network as an outsider? It's not a big network yet, so it may be easy to get started.
You could think of it in steps:
- Full network, but without edges. just who is on there. Mostly DHT scan stuff.
- Try and figure out edges. Could use things like gossip source and seq nr, and maybe other weaknesses / legacy things in libp2p gossip or streams that can tell you something. Document it, and we have things to look at for spec improvements :)
- Try and figure out validators. You can try the intersection approach with shuffling and validators, document how that goes, etc.
jrhea commented
Things of value that come to mind (which you covered most in your doc):
- monitor (and visualize) live network health and metrics
- conduct attacks on testnets to observe behaviors
- eclipse attacks
- denanonymize and DoS proposers
- Fill network with tons of peers embedded in the p2p mesh then drop them all at once
- attempt to partition the network by spinning up agents and selectively gossiping depending on peer/message
- if the communications happen quickly enough with the server, you could also spin it into a redundancy service in which a single node has many lightweight views and publish points into the network