#dev 2025-04-15

2025-04-15 UTC
troojg, stefen2, sebsel_, hidjy[d]_, aaronpk_, _hidjy[d]_, claudine, sebbu2, okCiel, srushe, immibis, rossabaker, roxwize, capjamesg and rrix joined the channel
nemonical, cow_2001, zoglesby and athenaeryma joined the channel
#
aaronpk
some bot is pulling down avatars at like 500 per second
#
capjamesg
😬
#
aaronpk
from about 6 different IP addresses, but all coming from https://nerdy.dev/ as a referer
#
capjamesg
And it is pulling from webmention.io links for avatars.
#
capjamesg
That post got a lot of traction.
#
aaronpk
maybe it's not bots?
#
capjamesg
If it is the same six IPs, it feels like it would be?
#
capjamesg
Maybe a crawler gone wrong?
#
aaronpk
omg so every page load is pulling 400 photos https://media.aaronpk.com/2025/04/15064150-4934.png
#
aaronpk
so probably it's not bots, i'm just seeing only a few IPs because it floods the logs
[snarfed] joined the channel
#
[snarfed]
Ugh, sorry
#
[snarfed]
Why is http://wm.io serving avatars? Shouldn't those be from Mastodon/Bluesky?
#
aaronpk
no, it downloads a copy of them, resizes them to a consistent reasonable size, and serves those URLs instead
#
aaronpk
hotlinking them from elsewhere was a giant mess
#
[snarfed]
Oh wow interesting ok, didn't realize
#
aaronpk
they're actually hosted on s3, but I kept the URLs as webmention.io URLs and the server sends a 301 redirect
#
[snarfed]
(and agreed, definitely downsides of hotlinking)
#
aaronpk
probably what I should do is use a subdomain on webmention.io that's pointed directly at the s3 bucket instead
#
[tantek]
That image needs to go on /facepile πŸ˜†
nemonical joined the channel
#
aaronpk
alright, the API now returns avatar URLs on a subdomain which is pointed to a cloudfront hosted website in front of the s3 bucket, so future crawls of the api will mean people archive the new URLs
#
aaronpk
all so that my logs don't get filled with 301 redirect requests πŸ˜‚
#
aaronpk
omg the nerdy.dev site is already serving the new urls
#
aaronpk
it must be loading the webmention api on each pageload too πŸ˜‚
duanin2 joined the channel
#
aaronpk
haha yeah when i visit the page, i see a request to webmention.io from a Deno user agent, which means his server is pulling down the api request each time
#
aaronpk
well cool that means my server is out of the hot water for now
#
capjamesg
aaronpk++
#
Loqi
aaronpk has 57 karma in this channel over the last year (131 in all channels)
[schmarty] joined the channel
#
[schmarty]
aaronpk++
#
Loqi
aaronpk has 58 karma in this channel over the last year (132 in all channels)
#
Loqi
CDN has 1 karma over the last year
#
[schmarty]
i was pretty impressed with this video on making a simple resizing/reformatting image proxy in about 20 minutes in PHP/Laravel. https://www.youtube.com/watch?v=lq_YlAOoLT8
#
aaronpk
oh yeah this isn't just a resizing proxy, it archives the file based on the content hash, so if the file at the URL ever changes the new file gets a new archived URL
#
[schmarty]
heck yeah! i didn't mean to imply anything about wmio's image proxy, just ADHD-brain pulling up this related thing i watched yesterday. πŸ˜‚
#
[schmarty]
i absolutely want to set up something like you described for the indieweb webring
#
aaronpk
sometimes i am amazed that webmention.io still runs
#
[artlung]
aaronpk++ fast fixing! amazing indeed
#
Loqi
aaronpk has 59 karma in this channel over the last year (133 in all channels)
#
aaronpk
it's almost 13 years old now
jacensekai joined the channel
#
capjamesg
Wow, that is a lot of requestsΒ±!
barnaby and [KevinMarks] joined the channel
#
[KevinMarks]
I had to nuke http://svgshare.com after Google stopped honouring daily billing limits and I got a $900 hosting bill for some crawlers
ttybitnik joined the channel
[Murray], jjuran_, ttybitnik, thegreekgeek, [aciccarello], [Joschi_Kuphal], nemonical, thegreekgeek_, gRegor, [tantek] and [tw2113] joined the channel