@jerry Just curious, what are the server specs running infosec.exchange? This instance runs beautifully compared to some of the others.
@xDL heh. Yes.
2 AMD 5950 16c/32thread 128GB ram with nvme ssds running sidekiq jobs
1 AMD 5950 16c/32thread 128GB ram with nvme ssds running elasticsearch and redis
1 AMD 5950 16c/32thread 128GB ram with nvme ssds running Postgres DB
2 AMD 5950 16c/32thread 128GB ram with nvme ssds running PUMA (mastodon web app) and nginx (only one is currently powered on - other is being reworked
1 AMD Epyc 32c/64th 400GB ram with 10TB of nvme ssd acting as a s3 storage provider and another nginx/PUMA web front end
@jerry @xDL Daaaaaaamn ๐Ÿคฏ That's a heck of a stack. #MastoInstance #MastoAdmin
@mspsadmin @xDL itโ€™s finally keeping up - we had a major period of growth since Thursday and it absorbed the load without an issue.
@jerry @mspsadmin Great job calling the growth. It's insane to think how much those boxes are getting slammed with all of the new people (including myself).
@xDL @mspsadmin thatโ€™s what theyโ€™re here for. I wanted people to have a decent landing spot. And I really didnโ€™t want peopleโ€™s first impression of the fediverse be skewed by bad performance.
@jerry @mspsadmin Definitely the right call. I used Diaspora back in the day and the particular instance I was using slowed to a crawl after a small set of users. I'm sure it was due to a tiny box hosted in someone's basement.
@jerry @mspsadmin @xDL lots of the fediverse seems to be suffering under the load of new arrivals. Iโ€™m impressed how well your stack is doing.
@mspsadmin @jerry @xDL Yes, I wonder why he needs such ? What's the minimum for Mastodon ?
@jindiggs @mspsadmin @xDL at the time, we had 20,000 simultaneous connections hammering the environment. Since then, itโ€™s fallen considerably (3000 simultaneous or so) and the footprint is smaller now, though I am having to expand it a bit again because itโ€™s starting to lag again
@jindiggs @mspsadmin @xDL also, the minimum for mastodon is pretty small - can be run on a raspberry pi for a single user instance