2y ago

[Done] Server will be migrated (More power!)

So after we've extended the virtual cloud server twice, we're at the max for the current configuration. And with this crazy growth (almost 12k users!!) even now the server is more and more reaching capacity.

Therefore I decided to order a dedicated server. Same one as used for mastodon.world.

So the bad news... we will need some downtime. Hopefully, not too much. I will prepare the new server, copy (rsync) stuff over, stop Lemmy, do last rsync and change the DNS. If all goes well it would take maybe 10 minutes downtime, 30 at most. (With mastodon.world it took 20 minutes, mainly because of a typo :-) )

For those who would like to donate, to cover server costs, you can do so at our OpenCollective or Patreon

Thanks!

Update The server was migrated. It took around 4 minutes downtime. For those who asked, it now uses a dedicated server with a AMD EPYC 7502P 32 Cores "Rome" CPU and 128GB RAM. Should be enough for now.

I will be tuning the database a bit, so that should give some extra seconds of downtime, but just refresh and it's back. After that I'll investigate further to the cause of the slow posting. Thanks @veroxii@lemmy.world for assisting with that.

78 comments

Wow that was fast.

Thanks for the awesome work!

I'm not too familiar with Lemmy's codebase, but I am a devops engineer. Is the software written in any way to support horizontal scaling? If so, I'd be happy to consult/help to get the instance onto an autoscaling platform eventually.
- The code is open source on GitHub and the backend is written in Rust.
  I have no idea how it goes in terms of scaling…
  
  Apparently it's not ideal at Horizontal scaling (that's what I've picked up from reading stuff here, could be wrong)
- Doesn’t support HA or horizontal scaling yet from what I read. Unsure if kbin does. Probably would have to add support for horizontal scaling to have that auto scaling do anything.
  
  Yeah, that's what I was afraid of. Understandable though, since horizontal scaling/HA usually isn't a priority when developing a new application.

Came here from Reddit and I already love it so much more! :)

Is one donation method preferred over another? That is to say, is one cheaper than the other?

Awesome to see so much growth. I just joined yesterday and I guess I'm not alone!

Does it work on water now that it has MORE POWA?

Just curious, what sort of hardware is lemmy.world using/moving to? Wondering if there's a good way to predict load based on number of users.
- Yes. It’s called performance testing. Basically an engineer would need to setup test user transactions to simulate live traffic and load test the system to see how everything scales, where it breaks, etc. Then you can use the results of the tests to figure out how big of an instance you should use for your projected number of users.
  Jmeter, and locust.io are the two biggest open source performance test tools.
  The alternative is take a wild guess. See how the system behaves, and make adjustments in real time… like what @ruud@lemmy.world is currently doing.

Like many others, I came from Reddit and was initially hesitant to try it out, but I love this place so much! It really feels like the "worse" parts of Reddit have been skimmed off, and that definitely shows with how nice people seem here! Thank you so much!
- Truth is for me as someone who used Reddit for about the last 16 years, it very much feels like the early days of Reddit again.
  Which is a very good thing, because that's what I originally signed up for compared to a metric fuckton of karma farming spam bots.
  I just hope it gains enough traction to be sustainable in the long run, especially considering that it's relying on donations for funding, I believe?
  
  undefined> metric fuckton of karma farming spam bots.
  People are hard at work writing bots for lemmy so don't worry, you'll be able to enjoy your regular hogwash again really soon.
  Personally I think lemmy should go as far out of its way as possible to make bots in any and all forms just about impossible.
- how nice people seem here
  yes! I love the culture of this place so far

So, I just want to make sure I understand this as I am a new user from reddit. Instances are server based and cost money. Instances are Lemmy.World, Beebaw, Lemmy.Film, etc etc. These are all seperate hosted instances. Correct?
And donations would help pay for the server, ie lemmy.world?
- That is correct. I've signed up for monthly donations to help cover costs (as well as added tip to help the admins themselves).
- "Lemmy instances" are analogous to "email servers": your account is hosted on one of them, but you can communicate with people on other ones, because the servers know how to talk to each other.
  Expanding the capacity of the Lemmy service will involve both (1) more instances, and (2) more resources for existing instances.

Just donated $10! Appreciate all the work you all are doing to keep up with the growth.

Thankyou for everything!

Thanks for you work on this! What is the planned time for the outage?

Nice!

Thanks for everything you're doing. I signed up for Patreon to contribute!

@ruud@lemmy.world DM me if you need help setting up monitoring/alerting on server health. IRL I'm on an SRE team, so happy to help where I can!

Itd be cool to get donation flare!
- You can! ☑️ or ✅. The Patreon page mentioned that you're officially allowed to edit your username to add flair when you donate. I upgraded to $8/month specifically so I could add the flair, but then got cold feet about the idea. 😀
  
  Nice! Why the cold feet? Just curious.

What kind of server configuration are you guys running? A single instance?

I don't understand why a dedicated server is a good idea, when the only true way to scale is to use like Kubernetes or Docker and ECS Containers with scale?
Your just gonna run into more problems, you cannot vertically scale forever.

Thanks for all you do @ruud@lemmy.world. You’re doing great!

Just joined. Thank you so much for your effort!

More power! More power is good.

Thank you very much! 🥳

For less tech-savvy newbies (like me), in case there is some confusion affecting your urge to engage/donate... My friend gave me a great explanation:
Lemmy the platform is planet Earth
“Instances” like lemmy.world, lemmy.ml, beehaw.org, etc. are like the different countries on Earth
When someone signs up, the user picks one instance to be a part of, like how an Earthling becomes a citizen of a country
If you register at lemmy.world, that means your home instance/ “home country” is lemmy.world, but you can “travel” to lemmy.ml, another instance / “country”, to check out and subscribe to their community
When you subscribe to a different instance that’s not your home instance, you can still participate in their content, and other people will be able to see which instance / “country” you’re from
Each instance can have its own version of the same “subreddit”, so you can have a c/Memes in your home instance that is different from a c/Memes in another instance. But you can subscribe to both separately
c/[community name] is the naming convention used here I think like r/[subreddit name] on Reddit. If talking about a community in a different instance, it's c/[community name]@[instance name] so like c/memes@lemmy.ml
Donations will help with the cost of running lemmy.world only and not lemmy.ml, beehaw.org, etc.
Someone please correct any of this if any of it is wrong, I’ll happily edit
- Is there a way to view C/Memes in all instances at once in aggregate? I don't want to miss out on what other instances are doing.
  
  Not yet, although there is ongoing discussion about this
  https://github.com/LemmyNet/lemmy/issues/818

Hello, i still doesn't quite grasp about the concept of federation and about how fediverse works.
But does it means that one instance can only run from one server?
Say lemmy.world running on Server A lemmy.ml running on Server B
User can register on whichever they want and can see the post from server A and Server B
But when Server A reach maximum capacity, can Server A scale up or distribute the load to multiple instances?
How can we solve the issue of computing power when more and more users migrate to using this services
Thank you 😀
Sorry if its a dumb question, but the whole Federation concept is still new to me. I created multiple account to log in to beehaw, mastodon, lemmy.world, lemmy.ml at first because i dont know that with one user, i can see other communities from another instances
- Optimal would be if users would spread over many servers, instead of all coming to Lemmy.world. But most users don't fully understand the Federation concept so they think they need to register here so they can see local content?
  I think the current server can handle a lot of users. It's just the software that isn't ready for it.. but that will improve. If ever this server gets too small, next step would be to scale using Kubernetes, but also that requires the software to be better prepared for that.
  
  Hello, after reading all the comments, I realized that I share the same questions (sort of) with the others.
  Thank you for replying and clarify things
  Cheers Ruud. And thank you 😊

Performance is looking awesome, lemmy.world is responding very fast to community subscription requests and search is also very fast. My experience when using other instances was that search didn't work at all, hindering community discovery.
Thanks!
- This is how I understand it: a current limitation (feature?) Is that you can only search from your instance to other communities if someone from your instance has interacted with it. But if you use https://browse.feddit.de/ you can search across all instances. Then subscribe to it, or search the whole url in your own instances search. Once an instance interacts with another, now other people from your instance can search for it by simple name.
  
  Oh, so it is due to the larger userbase here! There is a larger chance that someone already subscribed to a community I am looking for.
  Still, when I was using another instance, subscribing to communities at lemmy.world was instantaneous while subbing to communities at beehaw.org or lemmy.ml often took more than one try.
  It also doesn't help that lemmy.ml where a lot of users migrated at first seems to be having issues right now.
  Also on jerboa searching for communities by url doesn't seem to be working.
  Hopefully the influx of new users and attention helps improving and ironing some issues like it happened with mastodon.
  
  So, mostly correct. Lemme clarify:
  If you do a URL search in the communities page (with all settings set to "All", even "Communities"), your instance will pull in a few of the latest posts and comments. Not anything too heavy, just enough to give you an idea of what's going on.
  The moment a single user on your instance subscribes, your instance will start pulling in everything from that community. If every instance pulled in every community from every other instance, the network would be very vulnerable to a botspam instance that goes up would crash everything. Much better for an instance to only pull in communities that people are interested in.

Went ahead and subbed on patreon. Hope that lemmy survives the growing pains and can develop some of the community that reddit had!
Also if there are any fellow former apollo users would def recommend checking out Mlem, its in testflight right now but seems to be working towards the experience that apollo gave on reddit.
- iOS only? Or also Android? Btw, you receive notifications on Jerboa? What do you use for Lemmy on Android?

Would be awesome if you create some group chat (e.g. Discord?) and add sysadmins/devops to it. Would be more than happy to assist, especially if you have questions or need opinions.
I've been working as sre/sysadmin/devops for the past ~5 years and ~9 years of (Arch) Linux user. More than 1K Arch Wiki edits over that period of time.
- Reddit is for Lemmy as Discord is for Spacebar (works with Discord) https://github.com/spacebarchat/spacebarchat

Anyone looking to host something big should check out bare metal hosting like Datapacket, Reliablesite, FDCServers, etc. Down side is total lack of handholding and other cloud features and the fact that you can't scale up without redeploying on a new box, but the upside is ridiculously cheap bandwidth. The bandwidth cost is by size of pipe, not gigabyte transferred, and pipes upwards of 10gbps are affordable.
OVH and Hetzner are also worth looking at but aren't quite as cheap bandwidth-wise.
- With Hetzner Dedicated server the bandwidth is free

Are you running this out of your home?
I have self hosted small things before, but I was always curious about lager stuff like this.
What are your internet speeds?
- Check the updated post. It's running on a dedicated server hosted by Hetzner. Specs are high-end: "AMD EPYC 7502P 32 Cores “Rome” CPU and 128GB RAM."
  
  Thanks for letting me know. I also have my work stuff on Hetzner. But I do not see "Hetzner" listed in this post.
  Have you ever hosted on Vultr? I need a server with less latency and Vultr seems to have servers in a good location for my needs.

What happens if millions of users migrate over?

I would prefer if lemmy.world and mastodon.world donations were NOT comingled.
I want to donate to lemmy.world but already contribute to my mastodon instance (sfba.social) as well as the mastodon project.

78 comments