handled corner case, restarting/unhealthy container (e4a6369)
Users can now enable/disable clearing metrics to keep and maintain time series for gauge metrics even if their status was not 1

Notes

Update your prometheus alerting rules for cxp container status to handle non 1 status to catch restarting/unhealthy containers that are still running.

Assets 2

30 May 12:55

github-actions

v1.2.0

a62f8d9

v1.2.0

1.2.0 (2025-05-30)

Features

major internal migration from Flask and Gunicorn to FastApi and Uvicorn (a062184)
removed global variable init_container_names and moved the logic to util.metrics module leveraging registered metric labels as initial containers
no more new docker client per request!
handed over the event loop and task scheduling management to FastApi + Uvicorn

Notes

No changes to Prometheus metrics endpoints, names, labels, or scrape behavior.
Existing Prometheus scrapers, dashboards, and alerting rules will continue to work as-is.
The internal implementation is now fully asynchronous with FastAPI, potentially improving concurrent scrape handling under heavy load.
Logging and startup messages will differ due to the new framework and ASGI server (uvicorn).
adjust the following settings for uvicorn as environment variables:
- HOST
- PORT
- WORKERS (Default : 3)
- LOG_LEVEL=(Default : warning)

⚠️ Breaking operational change: if your deployment or runtime environment specifically depends on Flask or Gunicorn, you'll need to adjust service definitions accordingly.

Assets 2