Over the next week or so I’m sure a lot of people are going to try spinning up Lemmy instances - I’ve certainly been looking at it.
Does anyone have any recommendations for a VPS provider / resource allocation?
From what I have read, it sounds like you’re going to want a host that focuses on storage / bandwidth (at least if you are allowing image upload), but maybe those of you already operating an instance have a different opinion?
You must log in or register to comment.
I run lemmy.world on a VPS at Hetzner. They are cheap and good. Storage: I now (after 11 days) have 2GB of images and 2GB of database.
Could you give us some insight into how much bandwidth up/down your instance is using? Thanks for posting your server specs. I’m tempted to self host an instance when I finally get a gigabit connection in the next few months.
Just installed Zabbix agent, so data just coming in. Bandwidth usage for Lemmy.world server
There’s some talks about out-wards federation issues in which messages on big instances such as lemmy.world aren’t federating outwards. Do you mind sharing the CPU and memory data from Zabbix? Also, do you have NTP configured on the server? I’ve seen a couple people, myself also included, seen “Header is expired” messages in our logs during federation, implying something is backed up or timing is off. I’m curious if the issues are server performance related, or time related, or something else all together.
Server is performing nicely. NTP (timesyncd) is running. I have no idea how to see federation queues. Please contact me on matrix @ruud:h-y-p-e-r.space if you want to troubleshoot together.
I’m brand new to the fediverse; tried to message you via Matrix using Element – @chiisana:matrix.org – not sure if I did it correctly or tried to invite you to a room on the Lemmy space. Hope we can some how connect :)
Not sure about seeing queue stats directly but the logs should be pretty noisy if there aren’t enough federation workers to process the queue efficiently.
There was info from the devs here: https://lemmy.ml/post/1216911
E: I see you already commented there, so you already know. It seems the value was raised to 160k by the lemmy.ml instance though
Thanks for notifying. I raised it to 200k now.
I’ve used digital ocean with minimal fuss. I should write a start to finish guide on getting Lemmy going with one of their premade droplets.
If you ever get around to that guide I’d love a link.
Very basic but enough to get started/ https://novakeith.net/2023/06/14/setting-up-lemmy-on-a-digital-ocean-droplet/
Hetzner as always :) Although you probably want to separate out media storage to a standalone S3 provider or even just Hetzner’s SMB storage box
I have 4-5 nodes for miscellaneous projects hosted on cloudfanatic (previously servercheap). Cheap, good customer service, with servers all over the US. Good uptime. No international servers though.
FYI for people looking for a fully managed hosting of Lemmy, we do have an offer starting at $10 including the hosting + management fees, more details here in our Introduction post: https://lemmy.ml/post/1206609
Or directly on our website: https://elest.io/open-source/lemmy
If you’re willing to do the full VPS route like it seems, check out Hetzner. In my experience, it has higher resources at a lower cost than other providers like DO.
One other option is the “Always Free” tier on Oracle Cloud. You get some potato EPYC instances and some Altera ARM ones that are quite nice.
There are people who have issues with their accounts getting banned with no recourse, but I’ve used OCI free for over a year with no issues (and run a Mastodon instance on some of the ARM stuff), and know a good number of people who have various services running on it with no issue long-term, so YMMV.
The price is right, though, and you should keep current backups regardless.
I had this long post typed out on Jeroba about why I wouldn’t recommend it, but maybe I hit the character limit? IDK. Anyway, the point I wanted to get across is that I’ve been down that road, and up until February it was going ok, but one should absolutely not trust the Oracle free tier for any service that should be reliable long term, they can and will take the VM down and take back that generous free tier allotment, and IIRC sometimes without any notice. In my case it was literally because my VM was under utilized. No option to downscale my instance, just a notification that they’re taking my allotment back and deleting the VM a few weeks before it happened.
If you’re looking for a smaller company check out m5hosting.com
