# EPF6 - Week 20

## Overview
- Attended the Lodestar Weekly Standup and [shared updates](https://github.com/ChainSafe/lodestar/wiki/Lodestar-Weekly-Standups-2025#epf-contributors-progress)
- Focused on debugging and curating simulation runs to ensure consistent results between the **Kurtosis-based** and **regular Lodestar** simulation frameworks.
- Worked on comparing output logs to inspect differences between the **Kurtosis Runner** and the **original Runner**.
- Investigated the **table reporter** of the simulation as it's tailored on the original Lodestar’s testing framework
- Removed the manual time delay in the Kurtosis Runner after observing **no significant improvement** in the final simulation report.

## Main points

### 1. Debugging Simulation Consistency

While running both the **regular Lodestar simulation** and the **Kurtosis-based simulation**, inconsistencies were observed in timing and summary reporting:
- Variations in slot timings between the two **Runners**.
- Occasional missing or misaligned summary reports from the Kurtosis simulation.

Investigations from the [previous week](https://hackmd.io/@IB26/SJ9Iufp0gx#2-Timing-Differences-Between-JobJobOption-and-Kurtosis) focused on identifying whether the discrepancies were due to:
- **Timing**: Differences in **Runner termination timing**, especially around manual delay handling.
- **Simulation report**: Incomplete integration of the **[Table Renderer](https://github.com/ChainSafe/lodestar/blob/unstable/packages/cli/test/utils/crucible/tableRenderer.ts)**.
*Since the main focus has been on integrating the Kurtosis Runner, the table renderer hasn’t been extensively tested yet after the integration, and its behavior still needs to be validated in the new setup*

**Outcome:**
Despite the additional delay, **no meaningful improvement** was observed in the final simulation report.
The timing inconsistencies persisted, and the summary data remained non-deterministic.
As a result, **the manual delay was removed**, and the focus shifted to investigating deeper causes within the reporting and event handling logic.

> [!TIP]Next steps
> The issue likely originates from **mismatched reporting** event emissions between the legacy and Kurtosis Runners rather than timing delays, meaning the root cause of the mismatch between the two simulations **is not** related to timing itself.
>
> <font color="#27F535"> **Investigation will continue on the Table Renderer**</font>

## Key Learnings

| Area | Insight |
|------|----------|
| **Debugging** |The introduced time delay did not resolve inconsistencies, timing is not the main cause of mismatched summaries between simulations. |
| **Reporting** | Mismatched event emissions between legacy and Kurtosis Runners likely lead to inconsistent reporting results.|
| **Correctness Validation** | Running both simulations side by side remains the most reliable approach for verifying parity. |

## Week 21 TODOs

- Continue testing and fine-tuning simulation stability.
- Continue comparing simulation summaries between Kurtosis and regular Lodestar setups: investigation on the Table Renderer component
- Polish and clean up the code, focusing on readability and maintainability.
- Proceed with next steps entirely based on team and mentor feedback once the simulation behavior is confirmed stable.
- Refine final Devconnect presentation, incorporating findings from the consistency analysis.