Migrate to flask (whois-dev)

[whym]

I have been preparing a flask version of the tool inspired by @wiki-ST47 's fork.

https://whois-dev.toolforge.org/ is the new version. I want someone who is running an automated tool to fetch data from this tool to test it. If it goes well, I think I can replace the current version with it.

The current version is CGI-based, which means that a python interpreter launches every time someone tries to access the tool. The flask version runs continuously on a python interpreter. This change might provide better stability and slightly better performance (but not much because the bottleneck is in whois requests which are not cached.)

[AntiCompositeNumber]

I can point AntiCompositeBot's ASNBlock task there. It runs from about 02:30 UTC - 05:00 UTC and 08:13 UTC - 08:30 UTC. The number of WHOIS requests varies between the two runs and over time, but typical volumes range from a few hundred to a thousand requests.

[whym]

@AntiCompositeNumber Thank you for testing. Please let me know if you find issues.

It runs from about 02:30 UTC - 05:00 UTC and 08:13 UTC - 08:30 UTC.

I collected the beginnings of down time since June, and it looks like your range does not coincide with it a lot.

[whym]

I made a few changes. One surface level change is that parametarized urls ("gateway.py?ip=...") now redirect to REST urls ("w/..."). 0b66298

[whym]

In the recent few days (or longer) it constantly went offline for unknown reasons. I quickly checked the logs but found nothing about the cause. The log indicated that it went offline (and auto-restarted presumably) multiple times per hour.

2025-06-27 10:38:31: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 10:39:31: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 10:41:08: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 10:43:17: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 10:46:34: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 10:52:14: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 10:57:48: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:03:40: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:09:34: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:15:19: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:20:45: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:26:33: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:32:16: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:38:02: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:43:46: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:49:25: server.c.1513) server started (lighttpd/1.4.59)
2025-06-27 11:55:06: server.c.1513) server started (lighttpd/1.4.59)

As a temporary measure, I routed all traffic to whois-dev via a redirect.

I'll have to think about it more, but using this occasion to abandon the old version to move on to the Flask-based whois-dev might be a good idea, if it works stably.

[whym]

I might need to have a custom container image to use gunicorn to control traffic, if there is no way to do that with the ready-made webservice container.

EDIT: instead try configuring via uwsgi.ini

[gstrauss]

Please also consider using the latest lighttpd release, currently lighttpd 1.4.79, which is faster, uses less memory, and fixes bugs in lighttpd 1.4.59 (which was released 4 1/2 years ago and is old).