Introduction

Thanks to pytest-monitor, you can check for resource consumption of your tests written using pytest. However, experience in analyzing collected data turned out to be tedious:

  • it often requires specific code and queries to fetch the data

  • focusing on specific features requires specific processing

  • pytest-monitor does not support parallelism.

These are the main reasons why the monitor-server-api has been written. The monitor-server-api provides two components:

  • the monitor-server: a REST server for storing and fetching data

$> # Send pytest-monitor analysis to a monitor-server
$> pytest --remote http://monitor.instance.org:8181

  • the monitor-api: an API designed to easily query and filter data either from either

    • a monitor-server through the provided REST api

    • a local pytest-monitor database.

import monitor as my

# Connect to remote server
srv = my.Monitor('https://somewhere.org', verify=True)
# Get all sessions with both conditions met
#  - python set as a tag (regardless of its value)
#  - pandas tag set to value 1.0.1
with srv.read_sessions(with_tags=['python', 'pandas'], restrict_flags=['3.6', '1.0.5']) as sessions:
    # Retrieve metrics
    my_metrics = srv.get_metrics_from(sessions=sessions)
# Dump them as pandas DataFrame
print(my_metrics.to_df(keep=[my.COMPONENT, my.ITEM, my.KERNEL_TIME]).set_index(my.ITEM))