The locust stats outputs are already aggregated and don't calculate precise percentiles. We want to export every request made during the benchmark and then calculate 1 minute based hdr histograms, exactly like the Tideways backend does.
[x] Use JmeterListener from locust_plugins to get log output with every request tracked.
[x] Refactor LocustStatsParser to interpret Jmeter data.
[x] Add Histogram chart generation for overall, listings and product-details, including horizontal markers for median, 90%, 95% and include them in the report.
For the beginning we use the 95% percentile and group it into 5 grades A-D, F. We render a histogram with the distribution of responses into each grade category and compute a color based on 95% of requests being within a category.
The locust stats outputs are already aggregated and don't calculate precise percentiles. We want to export every request made during the benchmark and then calculate 1 minute based hdr histograms, exactly like the Tideways backend does.
LocustStatsParser
to interpret Jmeter data.For the beginning we use the 95% percentile and group it into 5 grades A-D, F. We render a histogram with the distribution of responses into each grade category and compute a color based on 95% of requests being within a category.
Fixes #24