State of Project Segfault: May 2024
This month was wayy more active than usual, and we got a lot of useful fixes, and a few new services :). Enjoy!
New instance: Shoelace
We are now hosting a new privacy frontend that allows you to see posts and users on instagram threads (the fediverse clone nobody likes) in a private way, let's see if Meta is going to actively try to destroy the project like they did with the Facebook and Instagram frontends :P
Invidious EU instance finally fixed (after 6 months of breakages)
We noticed that google has been...Let's say DDoSing us with subscription/feed update requests ever since we did our multiple-container scaling. To remedy this, we made sure the feed requests are only made from one invidious container, and we ratelimited those endpoints to make sure the instance doesn't break in the future!
The archived synapse data is now deleted
We kept that in case someone wanted to get his or her info to transfer to another instance, but since we haven't gotten any request for well over 3 months we decided to delete it to free some space.
Do note that this means that our previous hack to let images continue to be resolved will not work if your server hasn't cached those images while it was running.
DNS should no longer be broken
We initially said that we were in a limbo in the previous post, but we decided to stay with gandi for GLUE records and IPv6 support and made the projectsegfau.lt (main) domain point to psf.lt name servers, in case gandi screw us over we won't lose the primary domain at least. And yes, we looked into alternatives and Gandi is the best for .lt domain names (even Netim told us that, props to them for being honest when we asked!)
Mails to/from Proton should no longer be broken
As a result of the recent DNS breakages, and some old TLS-related records that ought to have been removed earlier, emails from and to proton addresses had been temporarily broken, but has now been fixed, thanks to one of our users who had issues delivering their mail to us so after some back and forth we concluded the problem came from the aforementioned DNS issues we had.
We banned Alibaba ASNs due to their traffic abuse making our instance be slow or unresponsive
We used to not have any ratelimiting or bans, except on Nitter due to the limited amount of accounts. And as you know, we had enabled logging last month, after looking into some logs that have raised suspicions based on instance usage, we concluded that a lot of Alibaba IPs have been hammering our servers for a while, after banning them, the traffic went back as normal. This should not affect any normal users, but if you are a legitimate user of any of the blocked ranges, please contact us so we can create an exception.
We enabled proper ratelimiting to our Invidious and Redlib instances to prevent abuse
This means that any abuse will be blocked using fail2ban, if you abuse our services and have been banned you are not welcome here and you should use another instance for your scraping purposes (or at least distribute the load between them all and you shouldn't be able to hit the treshold for being IP banned). If you happen to have been banned (be it through a VPN or TOR) please email us with the IP and we will check if it have been used for legitimate purposes or not, we will reply back if you have been unbanned or not, repetitive infractions means you will be sent to spam for the future requests.
New instance: Priviblur
To close it off, we added a new instance called Priviblur! It's essentially a Tumblr privacy frontend that is really light and fast.
That is all for this month, It's pretty packed compared to other months but we have a lot more things to do that we will report next month, have a great day and see you next month!
-Midou