Gregory's ServerFair, if a single client is hitting it excessively then it should be cached and isnt good etiquette for sure. Not quite sure I'd call it a DDoS but still, its bad design.
|
10 comments
@Gargron @freemo @xorowl @jimpjorps itʼs for the 'community tabʼ like this and the requests are done client-side, custom server owners using FiveM can select an ap account they want to use  @Gargron @freemo @mkljczk @xorowl @jimpjorps or, put differently: Is it distributed? Yes. Does it potentially lead to denial of service (through resource exhastion)? Yes. Sounds about right.  I cant speak to Gargron's setup but I think most setups would be able to handle 3400 RPM on the outbox without even batting an eye. Also by that logic if too many people start using mastodon clients on their phone or desktop then that is a DDoS since enough of them are distributed and would lead to resource depletion.  @freemo @rysiek @Gargron @mkljczk @xorowl @jimpjorps According to https://github.com/citizenfx/fivem/pull/757#issuecomment-873238836 it was 4,000 requests *per second*. That would make most instances cry.  then I misread it, that is on the high side.. though depends how many users were doing it. I would imagine mastodon clients in general produce more requests per second than that collectively but we wouldnt call those a DDoS... I dunno we are arguing semantics though, does it even matter what we call it? @freemo @tek @rysiek @mkljczk @xorowl @jimpjorps For comparison, average mastodon.social traffic is 200 req/s. I don't think it matters what we call it though. In my view it's a denial-of-service when it impacts performance due to unintended use, though maybe you could expand that to intended use as well. While the individual endpoints are intended to be used, it is the frequency with which they are retrieved that is unintended. |
@freemo @mkljczk @xorowl @jimpjorps Again, it's distributed DoS because they release this software (a game mod) to end-users whose IPs are the ones hitting the endpoints. As far as I understand, anyway. I'm currently analyzing the log files to find out how many unique IPs the requests are coming from.