techiev2 / process-monitor

Periodic callback based process monitoring app

Monitor app

A minimal IOLoop PeriodicCallback based application for monitoring database connection status. Runs a Tornado application on port 9999 and periodically checks for the global database object's status every 500ms.

On success/failure, initiates a specific notifier action as specified in one of the helper methods.

From a python shell,

from app import run_api
run_api()

From the terminal, using a virtual environment for the app.

PY_VERSION=3.6 ./scripts/setup
PY_VERSION=3.6 ./scripts/run

Trigger methods

# Update this method to trigger a custom notification/processing flow
# on a monitored process' downtime.
def notify_monitor_error(monitor_error):
    """."""
    pass

# Update this method to trigger a custom notification/processing flow
# when the monitored process is back up
def notify_monitor_success():
    """."""
    pass

# Update this method to trigger a custom check for any specific process.
def run_periodic():
    """."""
    pass

process-monitor's Issues

Add time series information for process uptime/downtime stats

To have a historic information of when the process went down and returned back to service, add a low latency data store and a time series data to generate the history.

Setup script fallbacks - System installation flows

Add flows to install system packages from source instead of depending on a singular apt-get install

• A way to abstract installation to work with other OSs - either from source, or a fallback based run through OSs.

• A way to install the requested Python version if it is not found, based on the previous flow.

Configurable connectivity options

The database in the app is a local mongo database. Need to add ways to configure database connections and a way to plug and play with other databases/services as the monitoring source.

Notification back end for the app

As it stands, the app debug prints the monitored process failure/return states. A way to plug a notification back end (email/slack/other) needs to be provided through an easily consumable API.