Skip to content

Latest commit

 

History

History
143 lines (99 loc) · 4.58 KB

README.md

File metadata and controls

143 lines (99 loc) · 4.58 KB
Raw

pontos-data-ingestor

A data ingestor from MQTT to a PostgreSQL database developed for the PONTOS project

Prerequisites

  • A running PostgreSQL instance with a database/table setup according to the setup outlined in PONTOS-HUB

  • A running MQTT broker

  • A data publisher that publishes to the MQTT broker on a well-defined topic structure using JSON payloads.

Specifics

The data ingestor extracts data from a single MQTT subscription and inserts into a single, narrow table in a PostgreSQL instance. Specifically, the data fields that are needed to populate the database table are the following:

  • timestamp (PostgreSQL format: TIMESTAMPZ)
  • vessel_id (PostgreSQL format: TEXT)
  • parameter_id (PostgreSQL format: TEXT)
  • value (PostgreSQL format: TEXT)

These fields are extracted from the MQTT topic and message payload (which is expected to be a JSON key-value construct) using the following "keys":

  • timestamp (MQTT format: seconds since epoch)
  • vessel_id (MQTT format: str)
  • tag (MQTT format: str)
  • index (MQTT format: int)
  • value (MQTT format: int/float/str (will be casted to str))

Note: parameter_id will be constructed through concatenation of tag and index as parameter_id = tag_index.

The data ingestor provides a flexible configuration setup of where to find these "keys" in the topic and message payload respectively through two environment variables:

  • TOPIC_PARSER_FORMAT

    A format string adhering to the principles of the parse library.

    Example:

    TOPIC_PARSER_FORMAT="PONTOS/{vessel_id:w}/{tag:w}/{index:d}"

    Which will accept, for example, topics like PONTOS/imo_8602713/anemometer_twd_deg/1 and parse it as:

    {
  "vessel_id": "imo_8602713",
  "parameter_id": "anemometer_twd_deg_1"
}

  • PAYLOAD_MAP_FORMAT

    A set of "keys" mapping to message payload "fields".

    Example: PAYLOAD_MAP_FORMAT=timestamp=epoch,value=sensor_value

    Which will accept JSON payloads of the format {"epoch": 1685078782, "sensor_value": 42, ...} and parse it as:

    {
  "timestamp": 1685078782,
  "value": 42
}

NOTE: It is up to the user of this service to make sure that the configuration provides a single unique way to determine all 5 "keys from the MQTT topic and message payload!!

Configuration

The following environment variables can be used for configuring the data ingestor:

MQTT_BROKER_HOST
Hostname or IP of MQTT broker to connect to (Required)

MQTT_BROKER_PORT
Port of MQTT broker to connect to (Optional, default: 1883)

MQTT_CLIENT_ID
Client id to use for MQTT connection (Optional, default: None)

MQTT_TRANSPORT
Underlying MQTT transport protocol, can be either 'tcp' or 'websocket' (Optional, default: 'tcp')

MQTT_TLS
Wether to use TLS when connecting to the MQTT broker (Optional, default: False)

MQTT_CLEAN_START
Wether to request a clean start or not when connecting to the broker (Optional, default: True)

MQTT_SESSION_EXPIRY_INTERVAL
Session expiry interval used for any session created by the broker (Optional, default: None)

MQTT_USER
Username to use when connecting to the MQTT broker (Optional, default: None)

MQTT_PASSWORD
Password to use when connecting to the MQTT broker (Optional, default: None)

MQTT_SUBSCRIBE_TOPIC
Topic or topic wildcard to subscribe to from the MQTT broker (Required)

MQTT_SUBSCRIBE_TOPIC_QOS
Quality of Service level to use when subscribing to MQTT_SUBSCRIBE_TOPIC (Optional, default: 0)

PG_CONNECTION_STRING
PostgreSQL connection string (Required)

PG_TABLE_NAME
Database table name to insert data into (Required)

PG_POOL_SIZE
Database connection pool size (Optional, default: 1)

TOPIC_PARSER_FORMAT
See explanation and examples above (Required)

PAYLOAD_MAP_FORMAT
See explanation and examples above (Required)

PARTITION_SIZE
Size of batches written to database if within PARTITION_TIMEOUT (Optional, default: 25)

PARTITION_TIMEOUT
Maximum time between writes to database (Optional, default: 5)

LOG_LEVEL
Log level of application (Optional, default: WARNING)

DISCARD_NULL_VALUES
Wether to discard null values or not in the parsed fields (Optional, default: false)

Development

The repository includes a devcontainer setup which is the recommended way of creating a development environment. See here for a generic get-started in VSCode.

Linting:

black .
pylint *.py

Testing:

pytest -v tests/

License

See LICENSE