badger-sarafu testing

[michielbdejong]

First I'm testing with the minimal loop (edited for #32):

6 4 0.5   -> nothing should happen yet
3 6 100.0 -> nothing should happen still
4 3 2.0   -> 4 should send 3 a flood probe
          -> 3 forwards the probe to 6
          -> 6 forwards the probe to 4
          -> 4 would detect the loop and send a trace to both 6
          -> 6 forwards the trace to 3
          -> 3 forwards the trace to 4
          -> 4 announces a loop

[michielbdejong]

The logging is now:

michiel-2:strategy-pit michiel$ npm run build ; node build/src/run.js

> [email protected] prebuild
> npm run lint


> [email protected] lint
> eslint . --ext .ts --ext .mts


> [email protected] build
> tsc -p tsconfig.json

This simulation will take about 60 seconds to complete.
Adding node 6
Adding node 4
Meeting "6" "4"
Adding node 3
Meeting "3" "6"

0: [6]->[4] meet
0: [6]->[4] probe 847c34c9
0: [3]->[6] meet
0: [3]->[6] probe 910d6f7f

1: [4]->[6] have-probes
1: [6]->[3] have-probes
1: [6]->[4] probe 910d6f7f

2: [6]->[4] okay-to-send-probes
2: [3]->[6] okay-to-send-probes

3: [6]->[3] probe 847c34c9

4: [3]->[6] have-probes

5: [6]->[3] okay-to-send-probes


Loops found:
3 []
4 []
6 []

I was a bit surprised that the probes are already created when a node meets its first contact, since in that situation the node can be sure to be on the edge of the network (i.e. a leaf and not part of a loop). I guess I can optimize this -> #32

[michielbdejong]

20 transactions now run in 100ms, yields 17 paths:

0.3.4.0
0.3.6.4.0
0.3.6.5.7.4.0
0.3.6.7.4.0
0.3.7.4.0
0.3.7.5.6.4.0
3.4.6.3
3.4.6.5.7.3
3.4.6.7.3
3.4.7.3
3.4.7.5.6.3
3.6.5.7.3
3.6.7.3
3.6.7.4.3
4.6.5.7.4
4.6.7.4
5.6.7.5

and this example graph for probe possible (blue = probe, green = trace, red = announce):

[michielbdejong]

With less logging, 250 transactions take 10 seconds to simulate on my laptop (still the same 17 paths found).

[michielbdejong]

with 1000 transactions it runs out of memory after 3:26 minutes

[michielbdejong]

Transactions 19-7932 are all DISBURSEMENT ("faucet") lines, from 0 to some (often new) node. So probably easy to optimise!