Closed return42 closed 1 year ago
Understandable. I'll just remove the periodic polling for SearXNG and update the nightly build to factor in instance availability when generating the instance list.
We can also increase the frequency of checking the health of each instance. @dalf proposed to do that, I'm not what's the current frequency (one per hour)?
The implementation of farside instances do a search request on each SearXNG instance to see if the instance is online and does it work.
From the view of a SearXNG instance, the sum of the farside instance behave like a bot net that has been gone wild / doing the same query:
https://github.com/benbusby/farside/blob/6048d3c23ea5bcc289747989f305c38a5b4b9ecb/config/config.exs#L16-L18
.. again and again, 24/7 from different IPs. In SearXNG we tend to block wild bots since they will have a negative impact on the engines of the SearXNG instance when we pass trough these requests.
My suggestion:
To avoid this flood of requests on each SearXNG instance, the farside instances can use the list from the SearXNG team to see what instances are online --> https://searx.space/
May the farside team has more / other suggestions to improve the situation / Thank you for your understanding.