HTTP Requests are still logged at LOGGER_LEVEL=ERROR

[nano9g]

Describe the bug
On a self-hosted install, when LOGGER_LEVEL is set to be less verbose than INFO, all incoming HTTP requests are still logged. (In other words, everything related to interacting with the ChangeDetection.io web interface.)

Version
v0.55.8

How did you install?
Docker

To Reproduce

1. Set LOGGER_LEVEL=ERROR on the ChangeDetection.io container environment
2. Visit the ChangeDetection.io interface in a browser
3. Observe that large amounts of HTTP logs are generated

Expected behavior
LOGGER_LEVEL=ERROR (or WARN) should also suppress HTTP request logging. A new variable to control HTTP log level would also work.

Screenshots
Here's a log excerpt:

changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET / HTTP/1.1" 200 -
changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET /static/favicons/site.webmanifest HTTP/1.1" 200 -
changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET /static/favicon/3ae23823-c4a3-4d74-bf19-5cec69657f51 HTTP/1.1" 200 -
changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET /static/images/spread.svg HTTP/1.1" 200 -
changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET /static/favicon/2461a2c0-d169-440f-8db5-086b976374a0 HTTP/1.1" 200 -
changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET /static/js/watch-overview.js HTTP/1.1" 200 -
changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET /static/favicon/b2ca4e7f-505a-4f40-bf2c-c3e91f428cf3 HTTP/1.1" 200 -
changedetection  | 172.16.2.130 - - [17/Jul/2026 07:37:19] "GET /static/favicon/b6c12e29-6830-4cef-80f1-ffd39edd425e HTTP/1.1" 200 -

Additional context
Please let me know if any additional information is needed. Thank you!

[ebarkhordar]

Reproduced on current main (71d332d). With LOGGER_LEVEL=ERROR, the werkzeug request log lines still come through:

$ LOGGER_LEVEL=ERROR python changedetection.py -C -d /tmp/ds -p 5000
...
127.0.0.1 - - [18/Jul/2026 15:23:54] "GET / HTTP/1.1" 200 -
127.0.0.1 - - [18/Jul/2026 15:23:54] "GET / HTTP/1.1" 200 -

Root cause: LOGGER_LEVEL only reaches loguru. main() applies it to the loguru sinks (changedetectionio/__init__.py:342-351), and it does tame one stdlib logger explicitly (pyppeteer.connection at :360-361), but nothing sets the stdlib werkzeug logger. The access lines are emitted by werkzeug's dev server, which routes them through logging.getLogger("werkzeug"). On first use werkzeug.serving._log sets that logger to INFO when its level is NOTSET and attaches its own StreamHandler, so it stays at INFO no matter what LOGGER_LEVEL is. The two logging systems are independent, which is why the loguru threshold never touches the request log.

One option, mirroring what the code already does for pyppeteer at :360-361, is to give the werkzeug logger the same threshold when LOGGER_LEVEL is set. I tried it as a one-liner before the server starts:

logging.getLogger("werkzeug").setLevel(logging.ERROR)

With that in place and LOGGER_LEVEL=ERROR, three requests still returned 200 (server serves normally) but no access lines were logged. Note LOGGER_LEVEL accepts loguru level names (TRACE/DEBUG/INFO/SUCCESS/WARNING/ERROR/CRITICAL) while the stdlib logger wants the standard names, so mapping SUCCESS to something stdlib understands (or just capping werkzeug at WARNING when LOGGER_LEVEL is above INFO) would need a small translation. Happy to open a PR if that direction looks right to you.