[PSA] Admins: Watch for the antiyanks troll and consider adjusting your rate limits

"Antiyanks" is back at it again and has switched tactics to spamming a massive number of comments in a short period of time. In addition to being annoying (and sad and pathetic), it's having a deleterious effect on performance and drowns out any discussions happening in those posts. That spam also federates as well as the eventual removals, so it's not limited to just the posts being targeted.

Looking at the site config for the home instance of the latest ~~two~~ three alts, the rate limits were all 99999999. 🤦‍♂️

Rate limits are a bit confusing, but they mean: X number of requests per Y seconds per IP address.

The comment API endpoint has its own, dedicated bucket. I don't recall the defaults, but they're probably higher than you need unless you're catering to VPN users who would share an IP.

Assuming your server config is correctly passing the client IP via the XFF header, 20 calls to the /create_comment endpoint per minute (60 seconds) per client IP should be sufficient for most cases, though feel free to adjust to your specific requirements.

Edit: A couple of instances accidentally set the "Messages" bucket too low. That bucket is a bit of a catch-all for API endpoints that don't fit a more specific bucket. You'll want to leave that one relatively high compared to the rest. It's named "Messages" but it covers far more than just DMs.

38 comments

Thanks for the heads up. I don't know what 'Antiyanks' is, but I already had to ban one comment spammer.
The rate limits are indeed a bit confusing. The settings are:
Rate Limit: X Per Second: Y
I understand this to be 'X for every Y seconds'
So, a 'Comments' Rate limit: 10, Per second: 60, means a maximum of 10 comments per minute, correct?
Maybe the reason you see 99999999 is due to troubleshooting. I have increased my instance's limits multiple times while troubleshooting server issues, because the meaning of the settings was not clear to me. These limits are usually not the reason for the sever issue, but I put some high number and did not bring them back down after the issues were resolved.
I have lowered them now to more reasonable numbers. I will also be more strict with new applications for the time being.
- Hmmm - after changing these settings to what I think are reasonable settings, the server crashed and I am now getting 'Too many requests' messages.... So, perhaps there is something not working so well with these rate limits, or I am still misunderstanding their meaning.
  
  Not sure. I had mine set to 20 per 60 for a long time without issue.
  Most likely cause would be the Lemmy API service not getting the correct client IP and seeing all API requests come from the reverse proxy's IP.
  Are you sending the client IP in the X-Forwarded-For header? Depending on how your inbound requests are routed, you may have to do that for every reverse proxy in the path.
  
  how did you fix?
- So, a 'Comments' Rate limit: 10, Per second: 60, means a maximum of 10 comments per minute, correct?
  Correct, per client IP.
  Maybe the reason you see 99999999 is due to troubleshooting
  Could be. I try not to speculate on "why" when I don't have access to the answer lol.
  I don't recall any of them being from mander (unless they were dealt with before I started testing?), but thanks for taking preventative measures :)
  I don't know what 'Antiyanks' is
  It's the codename for a particular long-term troll and is based off of their original username pattern (which they still use sometimes). I have reason to believe it's also the same troll that used to spam the racist stuff in Science Memes.
  These are most of today's batch (minus the JON333 which was just a garden-variety spammer that made it into the last screenshot).
  
  I don’t recall any of them being from mander (unless they were dealt with before I started testing?), but thanks for taking preventative measures :)
  undefined
  
  I don’t know what ‘Antiyanks’ is
  It’s the codename for a particular long-term troll and is based off of their original username pattern (which they still use sometimes). I have reason to believe it’s also the same troll that used to spam the racist stuff in Science Memes.
  These are most of today’s batch (minus the JON333 which was just a garden-variety spammer that made it into the last screenshot).
  No, they were not in mander.xyz. But I am generally quite relaxed when it comes to accepting applications. I mostly reject an applicant if it is very clear it is not an actual user, and then actively follow up on recent accounts for a short time. So the possibility of silent spammer accounts accumulating over time is always a concern.
  
  So, a ‘Comments’ Rate limit: 10, Per second: 60, means a maximum of 10 comments per minute, correct?
  Correct, per client IP.
  Setting the limits to more reasonable values, like '20 posts per minute', causes the server to stop serving posts. My front page goes blank.
  So, I am starting to think that '20 pots per minute' means 'requesting 20 posts per minute' and not 'creating 20 posts per minute'.
  I am still having doubts about what these limits mean, but setting reasonable numbers seems to break things, unfortunately.

Really bad thread-breaking comment spam under this post: https://lemmy.world/post/34824537
- I was wondering about that. The same comment got spammed like 800 times.

I don’t operate a fed server, but just curious. Couldn’t/shouldn’t rate limits be on a per-user/per-session-token basis to avoid the vpn issue you mentioned?
- You'll have to talk to the lemmy devs about that. I'm a retired admin, but last I was aware, they're based on client IP.
- And that would solve issues with the reverse proxy not forwarding the ip correctly right?

ok, I fucked up and now my instance is erroring :(
- If you have DB access, the values are in the local_site_rate_limit table. You’ll probably have to restart Lemmy’s API container to pick up any changes if you edit the values in the DB.
  100 per second is what I had in my configuration, but you may bump that up to 250 or more if your instance is larger.
  Fixed:
  UPDATE local_site_rate_limit SET message = 999, message_per_second = 999 WHERE local_site_id = 1;

Thanks for sharing

I'm not an admin or really THAT technically knowledgeable when it comes to underlying infrastructure of these things, but because you mention VPN users in reference to shared IPs - would it be worth considering and mentioning mobile users or users otherwise on CGNAT networks?
For example, TMobile Home Internet would result in multiple users being represented by a shared public IP. Maybe these exit nodes don't have nearly the number of users under one IP in comparison to users behind a popular VPN services assigned IPs? I don't know, but thought it might be relevant! I understand it's also a tool geared toward combatting this spam and only so much can be considered against the improvement.
- That's a consideration, yeah,, but they'd have to all be hitting lemmy.zip (your instance) and all from the same /32 IPv4 address.
  (AFAIK) CG-NAT still uses port address translation so there's an upper limit to the number of users behind one IP address. They also are distributed geographically. So everyone would need to be in the same area on the same instance to really have that be an issue.
  The more likely scenario would be multiple people in the same IPv4 household using the same instance. But 20 comments per minute, divided by two people in the house would still be 10 comments per minute. That's still probably more than they could reasonably do.
  Edit: You mentioned T-Mobile internet. T-Mobile is pretty much all IPv6 with IPv4 connectivity via CG-NAT. lemmy.zip is also reachable over IPv6, so in that situation,it would try IPv6 first and CG-NAT likely wouldn't even come into play.
  
  that makes sense! thanks for the break down, and good to know!

Seems some others (or maybe alts? I don’t have the server IP logs) are getting in on the action. Saw this a minute ago

38 comments