This exporter parses the output of the Gatling (v2.3.1) simulation.log file. It will report the response time for every operation under the loadgenerator metric using the status (which can be Success or Failure) and sim_id (indicating the simulation that's being executed). For the failed operations we'll report a count of the number of errors that occured grouped by the type of error and simulation id (the status label will indicate Failure for these aggregates in order to aid filtering) A loadgenerator_summary metric is also generated and it reports the number of successful and failed executed statements.

Example:

# HELP loadgenerator loadgenerator statements response time
# TYPE loadgenerator gauge
loadgenerator{status="Success",sim_id="Select_GroupBy_StrShort"} 4.0
# HELP loadgenerator loadgenerator statements response time
# TYPE loadgenerator gauge
loadgenerator{status="Success",sim_id="Select_GroupBy_StrShort"} 3.0
# HELP loadgenerator loadgenerator statements response time
# TYPE loadgenerator gauge
loadgenerator{status="Success",sim_id="Select_GroupBy_StrShort"} 3.0
# HELP loadgenerator loadgenerator statements response time
# TYPE loadgenerator gauge
loadgenerator{status="Success",sim_id="Insert"} 29.0
# HELP loadgenerator loadgenerator statements response time
# TYPE loadgenerator gauge
loadgenerator{status="Success",sim_id="Select_GroupBy_StrShort"} 2.0
# HELP loadgenerator_summary
# TYPE loadgenerator_summary gauge
loadgenerator_summary{status="Success",sim_id="Insert"} 91.0
loadgenerator{error="Can't assign requested address: localhost/127.0.0.1:4200",sim_id="Insert",status="Failure"} 1668.0
# HELP loadgenerator loadgenerator aggregated errors count
# TYPE loadgenerator gauge
loadgenerator{error="Can't assign requested address: localhost/127.0.0.1:4200",sim_id="GroupBy_Int",status="Failure"} 21.0
# HELP loadgenerator loadgenerator aggregated errors count
# TYPE loadgenerator gauge
loadgenerator{error="Can't assign requested address: localhost/127.0.0.1:4200",sim_id="GroupBy_Long",status="Failure"} 19.0
# HELP loadgenerator loadgenerator aggregated errors count
# TYPE loadgenerator gauge
loadgenerator{error="Can't assign requested address: localhost/127.0.0.1:4200",sim_id="GroupBy_StrShort",status="Failure"} 19.0
# HELP loadgenerator loadgenerator aggregated errors count
# TYPE loadgenerator gauge
loadgenerator{error="Can't assign requested address: localhost/127.0.0.1:4200",sim_id="GroupBy_StrLong",status="Failure"} 19.0
# HELP loadgenerator_summary
# TYPE loadgenerator_summary gauge
loadgenerator_summary{status="Success",sim_id="Select_GroupBy_StrShort"} 272.0
# HELP loadgenerator_summary
# TYPE loadgenerator_summary gauge
loadgenerator_summary{status="Failure",sim_id="Select_GroupBy_StrShort"} 0.0

Prometheus ingests new metrics by scraping http endpoints. The gatling exporter will start a http server (default port is 9102) for prometheus to scrape. When the exporter is started it will start processing the new entries in the simulation log file (tail -f), and store them in an internal buffer (the length of the buffer is configurable using --buffer_len and defaults to 30_000) until prometheus scrapes them. Old entries in the buffer will be replaced by new simulation entries if the metrics are not scraped for a long enough time, in order to respect the configured buffer length.

In order to start the exporter you'll need to provide the path towards the gatling simulation file using -s Eg.

python gatling_exporter.py -s /opt/gatling-charts-highcharts-bundle-2.3.1/results/cratedb-1532095744058/simulation.log