ES_HOST - Elasticsearch host (can be either like this "http://elasticsearch:9200", or with login and password delimited by : and separated from the host name by @)
ES_USER - Elasticsearch host login, set up here the username for elasticsearch, if you choose setup username here, in the ES_HOST you should leave only url without login and password
ES_PASSWORD - Elasticsearch host password, set up here the password for elasticsearch, if you choose setup the password here, in the ES_HOST you should leave only url without login and password
AMQP_URL - an url to the rabbitmq instance
AMQP_EXCHANGE_NAME - Exchange name for the module communication for this module, by default "analyzer"
LOGGING_LEVEL - logging level for the whole module, can be DEBUG, INFO, ERROR
GRAFANA_HOST - an url to the Grafana instance. NOTE if you don't want to see dashboards in Grafana, leave this environment variable empty.
ES_HOST_GRAFANA_DATASOURCE - an url to the Elasticsearch instance, which is accessible from Grafana image, and this url will be used for creating datasources for Grafana
POSTGRES_USER - postgres username to access the postgres database
POSTGRES_PASSWORD - postgres password to access the postgres database
POSTGRES_DB - the name of postgres database
POSTGRES_HOST - the host of postgres database location
POSTGRES_PORT - the port of postgres database location
ALLOWED_START_TIME - allowed start time for gathering metrics, default "22:00"
ALLOWED_END_TIME - allowed end time for gathering metrics, default "08:00"
MAX_DAYS_STORE - max days to store metrics, the metrics gatherer will delete data points which earlier than max days to store from today, default 500
TZ - time zone, it will let better understand allowed start and end time. default "Europe/Minsk"
TIME_INTERVAL - time intervsl for calculating metrics, available options "hour", "minute", "day"
ES_VERIFY_CERTS - turn on SSL certificates verification, by default "false"
ES_USE_SSL - turn on SSL, by default "false"
ES_SSL_SHOW_WARN - show warning on SSL certificates verification, by default "false"
ES_CA_CERT - provide a path to CA certs on disk, by default ""
ES_CLIENT_CERT - PEM formatted SSL client certificate, by default ""
ES_CLIENT_KEY - PEM formatted SSL client key, by default ""
ES_TURN_OFF_SSL_VERIFICATION - by default "false". Turn off ssl verification via using RequestsHttpConnection class instead of Urllib3HttpConnection class.
ES_PROJECT_INDEX_PREFIX - by default "", the prefix which is added to the created for each project indices. Our index name is the project id, so if it is 34, then the index "34" will be created. If you set ES_PROJECT_INDEX_PREFIX="rp_", then "rp_34" index will be created. We create several other indices which are sharable between projects, and this perfix won't influence them: rp_aa_stats, rp_stats, rp_model_train_stats, rp_done_tasks, rp_suggestions_info_metrics. NOTE: This prefix should be the same as for service-auto-analyzer image, this will ensure we check the same indices.
AUTO_ANALYSIS_MODEL_REMOVE_POLICY - by default "f1-score<=80|percent_not_found_aa>70", the conditions for removing custom auto-analysis models, so that they were retrained. The conditions are checked and applied if at least one condition is satisfied. The available metrics: f1-score, accuracy, percent_not_found_aa. The available comparison operators: >=, <=, <, >, =, ==.
SUGGEST_MODEL_REMOVE_POLICY - by default "reciprocalRank<=80|notFoundResults>70", the conditions for removing custom suggestion models, so that they were retrained. The conditions are checked and applied if at least one condition is satisfied. The available metrics: reciprocalRank, notFoundResults. The available comparison operators: >=, <=, <, >, =, ==.
METRICS_HTTP_PORT - by default "5000", the http port for checking status of the metrics-gatherer. It is used when you run the metrics-gatherer without Docker and uwsgi. If you use Docker, you will use the port 5000 and remap it to the port you want. If you use wsqgi for running the metrics-gatherer, you can remap the port with --http :5000 parameter in cmd or app.ini.
METRICS_FILE_LOGGING_PATH - by default "/tmp/config.log", the file for logging what's happenning with the metrics-gatherer.
Install python with the version 3.7.4. (it is the version on which the service was developed, but it should work on the versions starting from 3.6).
Perform next steps inside source directory of the analyzer.
- Create a virtual environment with any name (in the example /venv)
python -m venv /venv- Install python libraries
/venv/bin/pip install --no-cache-dir -r requirements.txt
- Activate the virtual environment
/venv/bin/activate
- Start the uwsgi server, you can change properties, such as the workers quantity for running the metrics gatherer in the several processes
/venv/bin/uwsgi --workers 1 --socket :3031 --wsgi-file main.py --master --http :5000 --threads 1 --lazy-apps --wsgi-env-behavior holy --virtualenv /venv