We are back after a major server migration! This is the first time we've upgraded our servers since VT Social was first conceived, and hopefully things should be much smoother *and* safer now!
VT Social is now running on a 3-server Proxmox cluster, with redundant Ceph storage for all data. The database, metadata, application, and all miscellaneous data are on NVMe storage, while local media and cached media from other servers are on HDDs. All data is 3x replicated. If the main server goes down, everything should automatically fall back to one of the backups with only a couple minutes of downtime.
In addition, we now have streaming DB replication, which means that if something horrible happens to the database, we can recover with almost no data loss, and we can even go back in time to any point we want within the last 48h. On top of that, we are upgrading our off-site backup sync frequency from daily to hourly (for both database *and* media).
While the main instance is still running on a single machine right now, this gives us plenty of room to expand and bring up more frontends on the other servers (or seamlessly add whole new servers), so we can easily handle more peak traffic if we need to.
We're keeping the existing Mastodon version for a few days, but we have a major upgrade cooking that will bring us the brand new, fancy, upstream quoting support. Stay tuned! ✨️