logdna / agent-benchmarks Goto Github PK

Currently, only memory usage time series is automatically generated.

We should also include a CPU histogram based on the data we already gather.

We should support saving results to S3 as a way to archive it and to allow Jenkins to retrieve them.

We should generate with gnuplot or any other tool the generation of png to make the results more human friendly :)

We should support running it twice with different commit hashes, one as a baseline and the other as comparison, to facilitate regression testing.

The amount of data generated and the length of the each execution is currently fixed:

We should add profiles that group these settings, for example:

• 5m-high-load: 5 minutes run w/ high load.
• 2h-med-load: 2 hours run with medium load (endurance).
• ...

Use eslint-config-logdna configuration and pre hooks.

We should add functionality to the endurance tests to stress retries making the ingester down for small intervals of time.