You can park here if you’re disabled, old, a parent, or expecting a wifi router.

Initial inspection, but not yet going inside.

Now we wait for them to get curious. I’m going to need another camera, aren’t I…

Something for the cats to enjoy this winter…

I am being watched. #CatsOfMastodon

LOOK AT THIS CUTE CHONK! 😻 #CatsOfMastodon

...and some silly photos #CatsOfMastodon

Got some good photos of the cats this afternoon... #CatsOfMastodon

The hiding spot. #CatsOfMastodon

Someone is not happy about the current situation. #CatsOfMastodon

Those curled paws 😻

#特に意味をなさない猫の計測 #CatsOfMastodon #猫

Today’s next problem - dealing with the question “is it dinner time yet?” #CatsOfMastodon

Interesting, but not unsurprising. Configured a managed challenge rule for those conditions, and there's currently a 0% success rate. Will keep an eye on the log and Sidekiq dashboards to see if there's a drop in traffic and server load with the new rule.

@shlee have you dealt with this kind of scraping before?

Straight up user agent blocking doesn't work here, because unlike the BS from a few days ago, these guys aren't being honest.

It makes sense that scrapers would go for those endpoints - after all, you can get a fair bit of public post data from them, not just on our server, but across the entire fediverse (note - this is only *public* posts)

Hmm. I wonder if it's possible to restrict that to logged in users only.

One nice thing about storing NGINX logs in json is that it makes some powerful analysis possible using simple tools such as jq. Doing a bit of filtering to only the 429 requests reveals a couple of interesting things.

1. These are all from what appears to be residential IPs, mostly in China but with scattered hits from Russia, Ukraine, the US and others.
2. They're all using the same generic user agent, imitating Chrome.
3. They're all requesting specific URLs - mostly the tags API

Doing a bit of Mastodon server log investigation today as this new dashboard is showing some interesting data. What jumped out at me here is the number of 429 (too many requests) responses. This only happens when there's more than 300 requests in 5 minutes, which means something's hitting the server pretty hard (like a scraper), but we've got AI scraping protection turned on at the CDN. Interesting.

Mi-tan being way too cute. #CatsOfMastodon