grafoleancollector 0.0.8

Common utilities for creating (controlled) collectors for Grafolean

About Grafolean Collector Python library

This is a Python 3 library which helps build data collectors (bots) for Grafolean, an easy to use generic monitoring system. It only comes handy for controlled (that is: non-custom) bots, which are managed through Grafolean UI. Examples of such bots are SNMP, ICMP Ping and NetFlow bots, which all use this library.

License

License is Commons Clause license (on top of Apache 2.0) - source is available, you can use it for free (commercially too), modify and share, but you can't sell it to third parties. See LICENSE.md for details.

If in doubt, please open an issue to get further clarification.

Installing

$ pip install grafoleancollector

Usage

Library grafoleancollector provides a framework for easier interaction with Grafolean backend API. It is not needed, everything can be done with calls to the API, but it does provide abstractions that should make a job of writing a bot easier.

An underlying assumption is that a bot caters for exactly one protocol, and that data is polled. If the data should be pushed then there is no need for a framework - simply publish the data to Grafolean when available.

This library provides a class Collector. It is expected that bot implementators will subclass Collector and implement missing functions. Class provides:

• fetch_job_configs() - a function for fetching "job configs" - for each applicable account, each applicable entity, and each applicable sensors (along with all the necessary details)
• execute() - a blocking function that performs periodic calls to jobs() and schedules the returned (periodic) jobs

The responsibility of developer is to:

1. implement jobs() function
2. implement a function that will get called whenever a job should be run (perform_job in example below, do_snmp in SNMP Bot). This function should call send_results_to_grafolean() to post results to Grafolean.

The corresponding changes in Grafolean frontend need to be made as well (support for the protocol - credentials, sensors, possibly another entity type). Currently this can only be done by modifying Grafolean frontend source code.

Implementing jobs()

The main purpose of this method is to split information about what needs to be done (usually this information is a result of calling self.fetch_job_configs()) into separate jobs. The way this is done is protocol-specific. For example, SNMP Bot needs to know about all the sensors on an entity in a single job, because it might be able to optimize queries (merge them, use BULK). On the other hand NetFlow Bot only handles a single sensor per job, because it doesn't need to merge them - which simplifies implementation.

A short example (which works with a fictional MyProtocol protocol) would look like this:

from grafoleancollector import Collector, send_results_to_grafolean

class MyProtocolBot(Collector):

    def jobs(self):
        for entity_info in self.fetch_job_configs('myprotocol'):
            for sensor_info in entity_info["sensors"]:
                # The job could be triggered at different intervals - it is triggered when at least one of the specified intervals matches.
                intervals = [sensor_info["interval"]]
                # `job_id` must be a unique, permanent identifier of a job. When the job_id changes, the job will be rescheduled - so make sure it is something that
                # identifies this particular job.
                job_id = str(sensor_info["sensor_id"])
                # Prepare parameters that will be passed to `perform_job()` whenever the job is being run:
                # (don't forget to pass backend_url and bot_token!)
                job_params = { **sensor_info, "entity_info": entity_info, "backend_url": self.backend_url, "bot_token": self.bot_token }
                yield job_id, intervals, MyProtocolBot.perform_job, job_params

    # This method is called whenever the job needs to be done. It gets the parameters and performs fetching of data.
    @staticmethod
    def perform_job(affecting_intervals, **job_params):
        # affecting_intervals: the intervals (subset of intervals yielded by jobs() method) which caused this job to be
        # triggered. Only useful if there is more than one interval that could trigger the job.

        # ... fetch data using `job_params` ...

        # send the data to Grafolean:
        send_results_to_grafolean(
            job_params['backend_url'],
            job_params['bot_token'],
            job_params['entity_info']['account_id'],
            values,  # dict; keys are paths (strings), values are corresponding values (numbers)
        )

backend_url = os.environ.get('BACKEND_URL')
bot_token = os.environ.get('BOT_TOKEN')
jobs_refresh_interval = 60

b = MyProtocolBot(backend_url, bot_token, jobs_refresh_interval)
b.execute()  # blocking

Development

Contributing

To contribute to this repository, CLA needs to be signed. Please open an issue about the problem you are facing before submitting a pull request.

Issues

If you encounter any problems installing or running the software, please let us know in the issues.