Its been a little while since I posted stuff :)
CPU:
Memory:
Network:
Storage:
Cloudflare caching:
Comments:
Not much has changed in quite a while. I still have a cron-job running to restart Lemmy every day due to memory leaks, hopefully this improves with future updates. Outside of that, CPU, memory and network usage are fine.
Object storage usage is growing steadily, but we’re a long way from paying more than the monthly minimum Wasabi fee.
I’m a bit confused by the upside-down negative swap graph. Does it say that the server is eating up to 5GB of swap, but restarting lemmy each night only returns that to about 2.5GB of swap? What’s the swap monster?
Yep, thats what it means. What is the swap monster? Seems to be a mix of the lemmy backend and the associated Postgres DB. Its likely the Postgres config is sub optimal and my fault, but honestly given the other issues the lemmy code has had in the past I’m happy to leave the daily restart in place.
“Get out ‘o me swap!”
Thanks for sharing! That sawtooth pattern on the memory graph is pretty crazy, it’s leaking like a sieve.
How are you going for server costs? Shout out if you need more donations, I think many of us have realised this is a viable platform for the long-term.
Current kitty is enough for the next ~7 months, thanks to some generous larger donations in the last few days. Ongoing monthly contributions currently cover about 50% of the monthly operating costs. Early on monthly donations exceeded monthly costs… leading to the current surplus. Between the surplus, and the existing monthly donations, we won’t hit the red for… 12 months?
Total costs are ~$85AUD per month for the server and object storage. I can cover that myself if needed.
tldr; finances are fine 🙂
Does CloudFlare caching help with bandwidth from TPAs, or does CF only cache front end web UI stuff?
TPAs?
CF is best for static content such as images, by default that’s almost all it caches. We’ve had some popular images cached and save us terabytes of egress traffic.
