.. _summary:

Summary
=======

NetSpyGlass comes with embedded Python interpreter that is used to process
collected monitoring data. At the end of each polling cycle, the server scans
Python scripts found in the directory `scripts/aggregation`, looking for those that
declare a class based on `nw2rules.Nw2Rules` with a method `execute()`. The server
loads all scripts like that and then calls method `execute()`.

Scripts are executed in parallel, sometimes by different servers, and do not
share memory. Scripts can not share variables directly with each other.

These scripts operate on monitoring variables and can modify them, create new
variables, create log records, add or modify tags on monitoring variables and
trigger alerts.

Here is an example of a simple aggregation script that uses NsgQL query to
find and count active agents and then exports the value it calculates as new
monitoring variable with name `numAgents`. The script can be checked in to git
as file `aggregation/count_agents.py`::

    import nw2rules
    import nw2functions


    class Aggregation(nw2rules.Nw2Rules):
        def __init__(self, log):
            super(Aggregation, self).__init__(log)

        def execute(self):
            """
            count customer's agents (those that do not belong to region "netspyglass")

            the agent is assumed to be "down" if we dd not receive its status for the last 5
            minutes. Export result as new variable numAgents
            """
            agents_query = \
                "FROM processUptime WHERE (NsgRole = 'agent' and NsgRegion NOT NULL and NsgRegion != 'netspyglass')"

            agents_count = sum(1 for mv in nw2functions.aggregation_query(agents_query, 300)
                               if mv.timeseries and mv.timeseries.getLastValue() > 0)

            aggr = nw2functions.new_var('nsg-agents', 'count')
            aggr.timeseries.put(nw2functions.current_timestamp(), agents_count)
            aggr.addTag('VariableTags.Aggregation')
            nw2functions.export_var('numAgents', [aggr])


The script defines class `Aggregation` (the name does not really matter) with method `execute()`.
Servers fund this script and call `execute()` on every cycle.