Post by @stefano
Dear #littleOne friends,
Current status: I've carried out a few measurements and changes. I fixed the backup issue, which was causing deadlocks in the queue until manual intervention. This morning it was stuck for about an hour and a half, resulting in a backlog of 1,900 jobs, which were cleared in less than an hour.
CPU load is very, very low, better than I expected. However, I occasionally notice some slowdowns that I suspect are due to the WASM driver currently used on NetBSD ARM (v6 and v7), which limits the number of database connections, even for read-only operations. I’m currently investigating this. But hey - do we really plan to host 40 users instances on such a low power device? (Yes, we do!)
The instance will likely return to the VPS soon, as soon as I finish looking into it, while we roll out further changes.
Stay tuned!
@shom@gts.shom.dev thanks. after all, keeping things online (as much as possible) while moving everything behind is my job 😃
@stefano @shom@gts.shom.dev but I doubt you're swapping between SBCs and VPSs at your job. If you are, that's super cool. And either way, this is an excellent scaling demonstration!
@evgandr@bsd.cafe