A great article about the Internet and #climate #change, about the web's comfortable walled gardens, urban elites, advertisements, the division between the rich and the poor, resilient communities, #decentralization and much more totally worth the 20 minutes of reading.
I found 1 assessment of "too much memory": https://github.com/wimvanderbauwhede/limited-systems/wiki/Pleroma-and-Mastodon-on-the-Raspberry-Pi-3
And 1 issue about ARM support generally: https://github.com/tootsuite/mastodon/issues/802
I run my mastodon instance on a Scaleway x86 server, with 2GB of RAM, 2 x86 cores and 20GB of storage.
It is currently using 1.2GB of RAM, a barely any load on the CPU and uh....most of it's disk space.
This instance has been up for about 4 months and runs very nicely for just me. Idk how many other instances its federated with, I would hazard at below 20.
I should add more disk. Thanks for making me check that @rixx !
@3rik @s @rixx @katrinleinweber 3 user instance here. We're prepared to take on more users (sign up is open). For now we're running mastodon with full text search enabled on a 2 core, 4gb ram dedicated server with 1tb storage.
The instance has been up for about 2 months and last time I checked it took up about 80gb of storage. Memory usage is at a constant 2gb or higher, most of which is elasticsearch. I think that is because it does in-memory caching + it runs on the jvm which loves memory.
It would be great to compare this with larger instances (@rixx ?) to also get a feeling if and how larger instances are more energy/ressource efficient per user or not - or even vice versa.
@3rik @hugot @s @katrinleinweber I'm pretty sure larger instances must be more enery/resource efficient, seeing as they don't duplicate a lot of the stored media files and data – if a thousand people from my instance follow you, your statuses, media files, profile pictures etc will only be saved once.
At the same time, I don't think this difference is on a scale that matters a lot.
I see the point. On the other hand, all clients of these thousand people will download the media files and data again from your instance and save it once more on their devices. But this will always be, not matter the size of an instance.
btw: I think we often underestimate how easy even little data transfers scale up.
@niklas has a very nice example of how 25kb scale up pretty high in just one month on #wikipedia: https://media.ccc.de/v/bub2018-2-warum_unser_web_nachhaltiger_werden_muss_und_wie_wir_das_anstellen#t=800
Server run by the main developers of the project It is not focused on any particular niche interest - everyone is welcome as long as you follow our code of conduct!