cross-posted from: https://lemmy.one/post/38848
If anyone here is running an instance on Lemmy, I’d like to know the system requirements recommended. I want to run an instance on a VPS, probably just as an account server.
What type of VPS can I get by on, for just accounts on an instance? How much storage space would I need to do just an “account instance”, and how much would I need for a full instance with maybe a few small communities?
lemmy.worldstarted on a 2vcpu/4GB instance which worked until it was like maybe about 1k registered users / 300 active users. It since scaled up to 4vcpu/16GB: https://lemmy.world/post/22070lemmy.mlran on 4CPU’s until it was like 30k registered and 1k active: https://lemmy.ml/comment/451028.sh.itjust.worksruns on pretty big hardware, but they actually posted resource utilization graphs: https://sh.itjust.works/post/4706. I think they were maybe 100 registered users and 20-50 active at the time this was originally posted (they’re one of the fastest growing instances and are much much bigger now). Seems like maybe 1G of ram usage?These are both big instances, I don’t know exactly what lemmy scales down to for a single-user instance if that’s what you’re after. But requires a postgres db and 3 other processes (lemmy, lemmy-ui, and pict-rs). It seems likely that 1CPU and between 512MB and 1GB of ram would be somewhere near the minimum resource requirements. Not sure about disk-usage, I haven’t seen anyone discuss it.
I have a single user instance, 1vCPU 1GB ram, and lemmy itself is using around 300MB of disk.
I’ve only had it for less than a week so maybe disk space will go up quite quickly.
CPU never goes above 10% while I’m using it and ram stays around 600MB.
The issue with disk are the logs, I solved it configuring docker to rotate files and just keep 4. Not sure how it could be done with the other installation instructions.
AFAIK the bottleneck currently is CPU, with the upgrade of lemmy.ml to the 8vCPU they are at full usage but ram sits around 6GB.
They’re working on optimizing some queries.
Awesome hardware report. If my instance starts falling over too hard I’m definitely considering setting up my own and this is great info to help me size it properly… though my admins lived through the mastodon twitterpocalypse with 4k signups per hour so if anyone can keep the lights on through this madness I’d think they have the experience to do it.
See I would like to help out if I could and open up my instance since I’m sitting on a lot of unused hardware, but I’m just not really interested in moderating.
Nothing wrong with that, admin’ing an instance is a pretty serious job. I wonder how many instances will fold once they realize what they really signed up for. I lucked out in finding a new Lemmy instance from an experienced mastodon admin team so I have some faith that they’re not in for a rude awakening.