prometheus-argus-glue

A service that receives alerts from Alertmanager (Prometheus) via webhooks and pushes events into Argus.

Some components and implementation is based off https://gitlab.sikt.no/cnaas/mist-argus

production deployment

gunicorn is recommended for production deployment:

$ gunicorn --bind 127.0.0.1:5000 promargus.webhook:app

Debug logging from the app is available if gunicorn is called with --log-level=debug.

configuration

See example_settings.py for configuration params.

ARGUS_SEVERITY_CONFIG is a dictionary with keys being a key, value tuple of the alert label, and value being the severity level. If set it enables the glue service to determine incident severity based of the alert labels. e.g. "labels": {"pagepolicy": "always"} could be mapped to severity 1 (Critical) like this:

ARGUS_SEVERITY_CONFIG = {
    ("pagepolicy", "always"): 1,
}

Note that the service will set the severity based on the first match, so order matters. For example, an event has the labels {"pagepolicy": "always"} and {"location": "WLG"}, if ARGUS_SEVERITY_CONFIG is:

ARGUS_SEVERITY_CONFIG = {
    ("pagepolicy", "always"): 1,
    ("location", "WLG"): 3,
}

The severity will be 1, whereas:

ARGUS_SEVERITY_CONFIG = {
    ("location", "WLG"): 3,
    ("pagepolicy", "always"): 1,
}

The severity will be 3.

developement/testing

Install conda env and activate:

$ make install
$ conda activate ./conda_env

Copy settings example and make changes:

$ cd promargus
$ cp example_settings.py local_settings.py
$ vi local_settings.py

Running dev server:

$ export PROM_ARGUS_SETTINGS=local_settings.py
$ python -m promargus.webhook

Format code:

$ black promargus

Lint:

$ flake8 promargus --max-line-length 88