A package to collect measurements published by EMA using MQTT and a SQlite database
Linux service to collect measurements pubished by EMA via MQTT. EMA stands for [Cristobal Garcia’s EMA Weather Station](http://www.observatorioremoto.com/emav2/meteoen.htm)
This service is designed to run in cooperation with my [ema Server also available in GitHub](https://github.com/astrorafael/ema).
emadb is a software package that collects measurements from one or several EMA wheather stations into a SQLite Database.
Desktop applicatons may query the database to generate reports and graphs using historic data. You can also monitor current station status.
Three data sources are available:
- Per hour minima and maxima values (historic data)
- individual samples (real time, 1 min. aprox)
- 5 min. individual samples (historic data) (NOT YET IMPLEMENTED)
Warning: Time handled by EMA is UTC, not local time.
## Installation & Configuration
The following components should be installed first:
- python 2.7.x (tested on Ubunti python 2.7.6 & Windows XP python 2.7.10)
- [Python PAHO MQTT module](https://pypi.python.org/pypi/paho-mqtt/1.1), downloadable via pip (Linux/Windows)
- [Python for Windows Extensions](http://sourceforge.net/projects/pywin32/). Must choose python27 builds. (tested on build 219, for python27 windows 32bits)
The Windows python 2.7 distro comes with the pip utility included. Open a CMD window and type:
pip install paho-mqtt
### Linux installation (Debian)
All executables are copied to /usr/local/bin
Type sudo emadb to start the service in foreground with console output
An available startup service script for Debian based systems is provided.
Type sudo service emad start to start it. Type sudo update-rc.d emad defaults to install it at boot time.
### Windows installation
NOT YET AVAILABLE
### EMA Server Configuation
By default, file /etc/emadb/config.ini provides the configuration options needed. This file is self explanatory. In special, the database file name and location is specified in this file.
### Reloadable Parameters
The emadb service supports on-line reconfiguration on a limited subset of parameters. This means that you will not loose incoming data while changing these parameters. The list is listed below:
[GENERIC] # If true, the EMADB server will hold incoming MQTT messages instead of # writting them to the SQL database # Useful to perform online database maintenance on_hold = no # component log level (DEBUG, INFO, WARNING, ERROR, CRITICAL, NOTSET) generic_log = INFO
[DBASE] # Full Database Path File Name dbase_file = emahistory.db # Directory where JSON data is located dbase_json_dir = config # Period for periodic task execution [minutes] dbase_period = 1 # Limit years (included) for the Date dimension (from Jan 1 to Dec 12) dbase_year_start = 2015 dbase_year_end = 2025 # Auto Purge RealTimeSamples table every day (at midnight UTC) # or let it grow dbase_purge = no
# component log level (VERBOSE, DEBUG, INFO, WARNING, ERROR, CRITICAL, NOTSET) dbase_log = DEBUG
[MQTT] # MQTT topics to subscribe # reconfigurable by reload mqtt_topics= EMA/+/history/minmax,EMA/+/current/status # component log level (VERBOSE, DEBUG, INFO, WARNING, ERROR, CRITICAL, NONSET) mqtt_log = INFO
Log file is usually placed under /var/log/emad.log. Default log level is INFO. It generates very litte logging at this level. File is rotated by the application itself. Two strategies are supported:
- Daily rotations at midnight. Useful if the service runs for a long time.
- Size-based rotation. Useful if service is starte/stopped several times during the day.
### Service start/stop/restart/reload
In Linux, this is done using the well-known init.d script with parameters:
sudo service emadb status sudo service emadb start sudo service emadb restart sudo service emadb reload sudo service emadb stop
The service emadb restart will stop and start the process, loosing the MQTT & database connections.
The service emadb reload will keep the MQTT connection intact.
### On hold mode
The service can be set to an on_hold mode where incoming messages are enqueued in RAM. instead of being written to the SQLite file. This can be useful to perform various database maintenance activities, whcih can include:
- Data Migration & clean up
To do so, set the on_hold flag in the config file to yes and reload.
### Updating the registered stations list ##
emadb will only insert incoming MQTT data if the EMA station is previously registered in the database. While you can update the database itself using SQL commands, the preferred approach is to edit the master dimension JSON files, usually stored in the /etc/emadb directory.
Edit the files using your favorite editor. Beware, JSON is picky with the syntax.
To append new data in these files, simply reload or restart the service. To modify existing data (i.e. changing longitude, latitude of existing stations), use the emadbload utility.
Type sudo emadbload -h to see the command line arguments.
### Real Time Data
The RealTimeSamples table is an aid for possible (more or less) real time monitoring of EMA weather stations.
Real time status messages are stored in this table. If the service runs in the background long enough, this table will be periodically purged approximately at 00:00:00 UTC to delete last days samples.
## Data Model
The data model follows the [dimensional modelling approach by Ralph Kimball] (https://en.wikipedia.org/wiki/Dimensional_modeling).
### Dimension Tables
- Date : preloaded for 10 years)
- Time : preloaded, minute resolution)
- Station: registered weather stations where to collect data
- Type : measurement types (Minima, Maxima, Samples )
- Units: an assorted collection of unit labels for reports
The Ùnits` table is what Dr. Kimball denotes as a junk dimension.
### Fact Tables
- MinMaxHistory : fact table contaning hourly minima and maxima measurements ffrom EMA weather stations.
- RealTimeSamples : fact table containing current EMA status messages.
CREATE TABLE IF NOT EXISTS Date ( date_id INTEGER PRIMARY KEY, sql_date TEXT, date TEXT, day INTEGER, day_year INTEGER, julian_day REAL, weekday TEXT, weekday_abbr TEXT, weekday_num INTEGER, month_num INTEGER, month TEXT, month_abbr TEXT, year INTEGER );
CREATE TABLE IF NOT EXISTS Time ( time_id INTEGER PRIMARY KEY, time TEXT, hour INTEGER, minute INTEGER, day_fraction REAL );
CREATE TABLE IF NOT EXISTS Station ( station_id INTEGER PRIMARY KEY, mqtt_id TEXT, name TEXT, owner TEXT, location TEXT, province TEXT, longitude REAL, longitude_text TEXT, latitude REAL, latitude_text TEXT, elevation REAL );
CREATE TABLE IF NOT EXISTS Units ( units_id INTEGER PRIMARY KEY, roof_relay TEXT, aux_relay TEXT, voltage TEXT, wet TEXT, cloudy TEXT, cal_pressure TEXT, abs_pressure TEXT, rain TEXT, irradiantion TEXT, visual_magnitude TEXT, frequency TEXT, temperature TEXT, relative_humidity TEXT, dew_point TEXT, wind_speed TEXT, wind_direction TEXT, lag TEXT );
- CREATE TABLE IF NOT EXISTS MinMaxHistory
( date_id INTEGER NOT NULL REFERENCES Date(date_id), time_id INTEGER NOT NULL REFERENCES Time(time_id), station_id INTEGER NOT NULL REFERENCES Station(station_id), type_id INTEGER NOT NULL REFERENCES Type(type_id), units_id INTEGER NOT NULL REFERENCES Units(units_id), voltage REAL, wet REAL, cloudy REAL, cal_pressure REAL, abs_pressure REAL, rain REAL, irradiantion REAL, visual_magnitude REAL, frequency REAL, temperature REAL, relative_humidity REAL, dew_point REAL, wind_speed REAL, wind_direction INTEGER, timestamp TEXT, PRIMARY KEY (date_id, time_id, station_id, type_id) );
CREATE TABLE IF NOT EXISTS RealTimeSamples ( date_id INTEGER NOT NULL REFERENCES Date(date_id), time_id INTEGER NOT NULL REFERENCES Time(time_id), station_id INTEGER NOT NULL REFERENCES Station(station_id), units_id INTEGER NOT NULL REFERENCES Units(units_id), voltage REAL, wet REAL, cloudy REAL, cal_pressure REAL, abs_pressure REAL, rain REAL, irradiantion REAL, visual_magnitude REAL, frequency REAL, temperature REAL, relative_humidity REAL, dew_point REAL, wind_speed REAL, wind_direction INTEGER, timestamp TEXT, lag INTEGER, PRIMARY KEY (date_id, time_id, station_id) );