Mon(itor)Eater

Program for consuming monitoring data, transforming them and uploading the results to InfluxDB.

The main program, moneater, reads the input on standard output and passes it to a custom parser called the eater. The eater is a, potentially user-written, Python class with a pre-defined interface. After parsing the input text line, the eater returns zero or more data points that will be then sent to InfluxDB.

There is also an alterante program, fileeater, that monitors the contents of log files. This is useful when the data is stored in a file instead of standard output. The fileeater can also monitor multiple log files.

Example usage:

program_with_output | moneater.py --host localhost --database mydb --table=program_data --tag location=berkeley --tag operator=kk ProgramEater

or

program_with_output | tee program.out
fileeater.py --host localhost --database mydb --table=program_data --tag location=berkeley --tag operator=kk program.out ProgramEater

Configuration

Database Settings

The database connection settings can either be specified as arguments or inside a file.

The use of arguments to specifci database authentication is not recommended. It will be visible to all users on a machine. It is only included for convenience.

The database configuration can also be stored inside a YAML file. It is passed to the MonEater programs using the --config/-c argument. It will be processed before the other database specific arguments.

An example of the format is:

db:
  host: localhost
  port: 8086
  user: root
  password: root
  database: mydb
tags:
  globaltag1: globalvalue1

FileEater Specification

The list of files/eaters used by a fileater program can be specified inside a YAML file. This also allows multiple files to be monitored by a single instance.

If the --spec/-s argument is present, then the file and eater are treated as paths to directory containing the log files and the specification file.

Each monitored file has its own eater, table and list of tags.

An example of the specicification format is:

monitor:
  - eater: test
    file: {logdir}/pb.out # The {logdir} will be replaced by `file` argument.
    table: test
    tags:
      tag1: value1

Eater Interface

The eater is a Python class with the following interface. It does not have to be called Eater.

class Eater:
  def __init__(self):
    """ Initialize any required tools or member variables """
    pass
  def parse_line(self,line):
    """ Parse line and return any data points for sending to InfluxDB.
    Return value:
    None - No new data will be uploaded
    Dictionary (or list of dictionaries) - points to be upload

	The dictionary can have either of of the following formats. The format
	is determined by checking for the special `fields` key.
	
	Format 1: key-value definition of a single point, multiple if returned as list of dictionaries
	In this case, the time is taken as the time when the input line was received.
	`{'field1':value1,'field2':value2}`

	Format 2: Also can be returned as a list for multiple points.
	`{'time':timestamp, 'fields': format1-definition-of-fields-for-point}`
    """
    # Parse input line into data
    return data

The name of the desired eater is passed to the moneater.py program in the command line as the path in a module. If you define a ProgramEater inside mypackage.py, then run moneater.py mypackage.ProgramEater.

Existing Eaters

The following eaters are shipped with MonEater.

• eaters.tabeater.TabEater Input format is tab-separate columns with the first row being the field names. For example:

Column1	Column2
0	0.123
1	0.5546
2	0.245