Eidolon (formerly VMTPsim)

https://doi.org/10.5281/zenodo.19229385

Discrete-event simulation framework for studying quorum-based consensus under extreme latency and intermittent disconnection.

This project now centers on Paxos and quorum-system design in topology-aware, high-latency environments (Earth, Moon, LEO, Mars).

Core thesis: Paxos safety depends on quorum intersection properties, not simplistic "majority voting." Quorum shape can encode physical topology.

• Paxos acceptor/proposer/learner simulation (paxos.py)
• Quorum system variants:
  • Majority
  • Flexible Paxos (q1 + q2 > n)
  • Grid quorum variants
  • Crumbling wall quorum (quorums.py)
• Topology-aware network model with location links and partitions (datacenter.py)
• Interplanetary progression demos (demo_step_1.py ... demo_step_9.py)
• Step 9 conjunction experiment:
  • Hard blackout model
  • Lagrange repeater refinement
  • CSV outputs and sweep/plot tooling

• Structured quorum semantics are enforced in the proposer path:
  • quorum checks use is_phase1_quorum / is_phase2_quorum
  • crumbling-wall phase constraints are active in execution
• Step 9 artifact pipeline is operational:
  • single-run CSV export
  • parameter sweep over Mars latency and blackout duration
  • SVG figure generation without external plotting dependencies

• Core simulation:
  • paxos.py
  • quorums.py
  • datacenter.py
  • network.py
• Interplanetary demos:
  • demo_step_7.py (hierarchical Earth/Mars)
  • demo_step_8.py (crumbling wall full topology)
  • demo_step_9.py (conjunction blackout vs repeater)
• Experiment tooling:
  • experiments/step9_sweep.py
  • experiments/plot_step9.py
  • experiments/step9_liveness.py
  • experiments/plot_step9_liveness.py
• Research docs:
  • docs/workshop-paper-roadmap.md
  • docs/step9-repro.md
  • docs/papers/224964.224978.pdf (Peleg & Wool)

uv sync

Run Step 9 baseline/repeater comparison:

uv run python demo_step_9.py \
  --mars-latency-s 186 \
  --blackout-start-s 600 \
  --blackout-duration-s 900 \
  --sim-end-s 3000 \
  --reconcile-interval-s 120 \
  --seed 42 \
  --csv results/step9/single_run.csv

Run the sweep and generate plots:

uv run python experiments/step9_sweep.py \
  --mars-latencies-s "186,750,1342" \
  --blackout-durations-s "300,900,1800" \
  --seeds "40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89" \
  --output results/step9/step9_sweep.csv \
  --aggregate-output results/step9/step9_sweep_ci.csv

uv run python experiments/plot_step9.py \
  --input results/step9/step9_sweep.csv \
  --output-dir results/step9/plots

uv run python experiments/step9_liveness.py \
  --mars-latency-s 186 \
  --timeout-s "120,240,360,500,720" \
  --blackout-durations-s "300,900,1800" \
  --seeds "40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89" \
  --output results/step9/step9_liveness.csv \
  --aggregate-output results/step9/step9_liveness_ci.csv

uv run python experiments/plot_step9_liveness.py \
  --input results/step9/step9_liveness_ci.csv \
  --output-dir results/step9/plots

• Sweep CSV: results/step9/step9_sweep.csv
• Sweep CI CSV: results/step9/step9_sweep_ci.csv
• Liveness CSV: results/step9/step9_liveness.csv
• Liveness CI CSV: results/step9/step9_liveness_ci.csv
• Single-run CSV: results/step9/single_run.csv
• Plots: results/step9/plots/*.svg

• RFC 1045: VMTP Specification (docs/rfc1045.txt)
• Peleg & Wool (1995): Crumbling Walls (docs/papers/224964.224978.pdf)
• Repro commands: docs/step9-repro.md