System drift analysis service
This is a flask app that provides an API for drift-frontend. It listens on port 8080 by default with gunicorn. Prometheus stats will be stored in a temp directory.
To set up pipenv:
yum install -y pipenv
pipenv install # will pull in deps and create virtualenv, and will print next steps to run
to set up temp directory (command will return a temp dir that you'll use later)
mktemp -d
To run (inventory service URL can be obtained from drift or platform team):
prometheus_multiproc_dir=mktemp -d INVENTORY_SVC_URL=<inventory service url> ./run_app.sh
To set the debug level (info by default):
LOG_LEVEL=debug prometheus_multiproc_dir=/tmp/tempdir INVENTORY_SVC_URL=<inventory service url> ./run_app.sh
The prometheus_multiproc_dir should be a path to a directory for sharing info between app processes. If the dir does not already exist, the app will create one.
The same info as above, but in handy table form:
| env var name | required? | expected values | description |
|---|---|---|---|
| INVENTORY_SVC_URL | yes | URL | URL for inventory service (do not include path) |
| LOG_LEVEL | no | string | lowercase log level (info by default) |
| prometheus_multiproc_dir | yes | string | path to dir for sharing stats between processes |
| PATH_PREFIX | no | string | API path prefix (default: /api) |
| APP_NAME | no | string | API app name (default: drift) |
If you would like to use this service with insights-proxy, you can use the
included local-drift-backend.js like so:
SPANDX_CONFIG=drift-backend/local-drift-backend.js bash insights-proxy/scripts/run.sh
If you use run_app.sh, drift app will be invoked via gunicorn. This should be
OK in most cases; even pdb runs fine inside of gunicorn.
However, if you want to use flask's server, use python3 standalone_flask_server.py
with the aforementioned environment vars.