Scenario Player Analysis

This is an analysis tool to show performance metrics using the logs of a scenario player run.

Prerequisites

• Python 3.7

We strongly recommend using a virtual python env like pyenv (in addition to that you can use virtualenvwrapper)

• Scenario-player

The Scenario-player and scenario-player logs are a prerequisite to using this tool.

Installation

pyenv shell 3.7.0
mkvenv raiden-analysis
pip install -r requirements.txt

How To Use

Either start the analysis directly or print out the help:

python analysis.py $PATH_TO_SCENARIO_LOG_FILE

python analysis.py -h

See an example result below.

Examples

Two different scenarios are currently available as examples and each have three generated files:

• A gantt plot is rendered in an HTML file - it contains a legend and a summary
• A CSV file is created which contains the duration of the tasks
• A HTML file with a statistic summary and histogram on the basis of each task type is getting calculated

Simple-Scenario

• Gantt: example/simple-scenario/gantt-overview.html
• CSV: example/simple-scenario/durations.csv
• Statistics: example/simple-scenario/statistics.html

Mediated-Transfers-Scenario

• Gantt: example/mediated-transfers-scenario/gantt-overview.html
• CSV: example/mediated-transfers-scenario/durations.csv
• Statistics: example/mediated-transfers-scenario/statistics.html

Limitations/Caveats

• the duration of a task is calculated by looking at the next event in the list of events (sorted by time)
• tasks processed in parallel may not be visualized correctly
• the number of ticks on the y-axis can be too high when too many tasks have been emitted in the log. In that case uncomment the showticklabels line within the draw_gantt() method.
• log messages are grouped by task-id, resulting in request/response calls being grouped