Grafana + something like influxdb+telegraf would do the trick. It sounds like you don’t have metric gathering like that on your instance? If that’s the case I’m surprised you don’t when you’re running with a full server rack haha.

Damn that setup is no joke. 21GB in a few months initially sounded like a lot to me… but I decided to math it out. Lets say the 20gb was across 1,2 or 3 months…

Time till 1tb would fill up.
+------+-----------+----------+----------+
|      | 3 months  | 2 months | 1 month  |
| 1 TB | ~12 years | ~8 years | ~4 years |
+------+-----------+----------+----------+

That data usage is looking pretty reasonable… Even 20gb per month is something that wouldn’t be too hard to keep up with and I’m sure eventually there’ll be a way to clean up old posts that no one on your instance saved or commented on if you are trying to save space. I’d start to worry if disk usage was hitting closer to 40gb a month.

Out of curiosity what has the disk usage growth looked like so far for your lemmy instance? I occasionally selfhost but I’m not a hardcore datahorder or anything so the replication of data from instances you subscribe to has me on the fence.