We are live on @upcloud@glitchy.upcloud.world servers! ![]()
Tonight, we moved mementomori.social from Hetzner to UpCloud servers in Helsinki. We had scheduled a maintenance window from 17:00 to 20:00, but our instance was down for just under an hour - the last post on the old server went out at 18:30, and the first on the new server went out at 19:30.
Some numbers from the move:
- PostgreSQL upgraded from 14 to 18; database size is 210 GB with about 118.5 million posts, 1.27 million known accounts, and 21.7 million media attachments
- 440041 posts written by our own 990 users
- Search: 34.9 million public posts, 1.24 million accounts, and 4.28 million hashtags indexed
- Data checksums enabled, so silent disk corruption now shows up as an error instead of bad data
- Rebuilt 72 text indexes for the new operating system's collation rules, ensuring ä and ö sort correctly. This also freed up about 20 GB
- Now accessible over IPv6 as well!
Every server is now provisioned with Ansible, and the playbooks are public: https://github.com/mementomori-social/ansibles
If your home timeline looks empty, it will repopulate as you use it. Please let us know if anything seems off.
Huge thanks to @rolle@mementomori.social and especially @ikkeT@mementomori.social for putting in countless hours. Thank you @upcloud@glitchy.upcloud.world for sponsoring our non-profit
Media files remain hosted on Cloudflare R2 as before.
We should now be faster, better, more sustainable, and more sovereign than ever. Party on!

