I was wondering why I was receiving a bunch of late replies...
Fediverse
A community to talk about the Fediverse and all it's related services using ActivityPub (Mastodon, Lemmy, KBin, etc).
If you wanted to get help with moderating your own community then head over to !moderators@lemmy.world!
Rules
- Posts must be on topic.
- Be respectful of others.
- Cite the sources used for graphs and other statistics.
- Follow the general Lemmy.world rules.
Learn more at these websites: Join The Fediverse Wiki, Fediverse.info, Wikipedia Page, The Federation Info (Stats), FediDB (Stats), Sub Rehab (Reddit Migration)
You're famous now!
WAWAWEWA
Wow sign my comment!
Good post though. Of all of them to get noticed, thats a good one :)
I signed it with an upvote. You're welcome.
Ill treasure it forever!
I wonder if the caching is not aggressive enough or something.
I had a PeerTube video from my instance on the HN FrontPage last week and the load was minimally higher compared to before or after.
I had several of my blogposts on HN FrontPage in the past. The first time it happened it brought my poor VPS to the Knies, but I learned from it and cached pages with nginx for some minutes and since then never had any problems. Just invalidate the cache when there are changes.
How does peertube cache? I know it P2P the video which is nice in case anything goes viral, but I never took the time to figure out any caching.
My tiny single digit user instance seems to keep up with any traffic on a equivalent of a pi.....so I assume its good .
I also put the video itself into a S3 bucket, so PeerTube basically only has to show the meta data and the comments from my server, so kind of like what Mastodon or Lemmy/PieFed has to do. I just had a look at the [PeerTube nginx config((https://github.com/Chocobozzz/PeerTube/blob/develop/support/nginx/peertube) but couldn't see anything there which would do caching, so I assume the app does it's own caching somewhere.
For my website, which is a rails application, I did
proxy_cache_path /var/lib/nginx/cache/jeena.net keys_zone=jeenanet:30m;
and then
location @rails {
# ...
proxy_cache jeenanet;
}
Nice! Yeah the S3 is a good way to work with it. They are dirt cheap at least.
...I should probably upgrade my setup at some point haha.
I had the problem that peertube redundancy only works on public videos and most of my videos are private/internal. And in my specific case I hosted them in Germany where my server is and because of routing and peering they would always buffer a lot in South Korea where I am so I had to solve it in a creative way, the S3 bucket is one part of my solution, putting it in the right country was another, which I explain in detail here: https://tube.jeena.net/w/uXZN52xsH75LbHWNt8dsLY
That's not a "post from feddit.org," that's a "post by a feddit.org user on a lemmy.ml community." It may be DDOSing feddit.org because that server's view of it is the link that happened to go viral, but there's no reason folks outside Lemmy couldn't view it on the server it's actually posted on, or some other third node like lemmy.world or whatever.
The original link ap_id
corresponds to the author's instance.
User@feddit.org posts to community@lemmy.ml. The path the post will take is this:
User -> Author's instance -> Community's instance -> Notified to all other instances subscribed, to download from author's instance.
As for your other point, there is no good universal Lemmy linking thing for posts. communities and users have names that work on any instance, but there's no similar unique identifier for posts.
As for your other point, there is no good universal Lemmy linking thing for posts.
In another comment I mentioned that there's this lemmyverse link thing. From the way you phrased that I'm not sure if you don't know about it, or do know but just don't think it's good, but I think it's about halfway to being useful.
What it needs to get the rest of the way is to:
- default to some kind of load-balanced or at least round-robin instance selection instead of defaulting to the original instance of the link given, and
- be built into the Lemmy UI(s) so that it's easy to grab a lemmyverse.link URL to share.
I have no idea how to get my app to open the original article ( https://feddit.org/post/18353777 ) so I'll just mention here that I installed PostmarkedOS on my ancient Wileyfox Swift a few days ago.
And while it was very exciting this particular setup was barely usable. I tried all the UI / DE options but only Phosh and KDE recognized touch inputs (so I could not do anything on Gnome and SXMO or whatever it is called). Phosh had some very weird problems with the onscreen keyboard and flickering menus and KDE had amazing graphical glitches (it really was quite something, I wish I had taken photos with another camera now) and sleep wasn't working even after the fixes from the wiki. I really wanna check it out on mever hardware at some point through, I think it's a really cool project:
DDoS by venture capitalism.
That was how the Internet worked back in the day. Site reliability for small sites was shit and going viral would routinely pull a smaller site down.
It's been somewhere on reddit as well today
Fashy tech bros incoming...
I'll be honest, there's a lot of reasonable people on HackerNews. Not only tech bros.
I would agree in the past but in the last few years this site has doubled down on the silicon valley brand of techno-fascism and I just can't read the comments without losing faith in "hacker culture" at this point.
Nice! I have a feeling that most of the fediverse dont use a whole lot of caching. I dont really enjoy some of the things cloudflare does, but it does do a good job with caching.
Is Memcache still a thing? I remember using that for a LOT of older apps back in the day.
IMO what we need is a different sort of thing more like load-balancing than caching, where users from outside the Fediverse (i.e. who aren't logged in to the server whose URL they're trying to load and thus probably don't actually have a strong preference about which instance they actually view it from) get redirected to other instances to help spread out the load.
There's no particular reason why the entirety of Hacker News and Reddit needs to be piling in to view this thing that was posted to !linux@lemmy.ml just because a feddit.org user happened to have been the one who wrote it (and possibly shared it).
Posts and comments already have a chain icon for the URL of the view of them on the user's current server and a fediverse graph icon for the URL of them on the poster's home server, but maybe there needs to be a third icon for the lemmyverse link url or something like that.
Very interesting idea, thanks.
Very interesting idea. Peertube already does this sorta. The p2p nature means the more traffic/servers, the less load the original instance has to push.
Wasn't that the idea behind zeronet? I think zeronet is dead(?) now, but it was a fun concept. https://github.com/HelloZeroNet/ZeroNet
Interesting, never knew!
I recall https://www.gnunet.org/ but its been a LONG time since ive played around with it.
I like I2p but it doesn't solve the problems that zeronet did.
Will have to look into gnunet though. Not sure what that does, or how.
Its basically i2p but older(?) and harder(?) to use. Its more like tor from what I understand.
https://docs.gnunet.org/latest/guis/gtk.html is a good place to start after the installation from what im seeing.
Sounds like a skill issue.