I have a self hosted Matrix server running on a VPS, with Debian 12, nginx, PostgreSQL 15.The matrix server became too large to host on the base mount point / so I added a storage block to the VPS and mounted that to /media then changed the PostgreSQL data directory to /media/postgres/main and symlinked /var/lib/matrix-synapse/media to /media/matrixmedia/media - so the PostgreSQL DB and the matrix media cache are sitting on the attached storage mount point and this all worked fine.Then my VPS host had some gigantic fuckup in their datacentre and a load of storage, including mine, went offline and had a bunch of data corruption when it was brought back up. When trying to start the PostgreSQL DB again it had "invalid primary checkpoint record", okay so I took a filesystem level backup and ran "pg_resetwal -f DATADIR" to reset the transaction log and it start successfully. Then I try to start matrix-synapse and it has the pic related error, so I run that SQL and it deletes 1 record then matrix starts successfully with no errors.But my server still appears to be down, all clients I'm currently logged in on still have "connectivity to the server has been lost", I try to log in on https://app.element.io/#/login and it just hangs on login syncing, but the federation tester https://federationtester.matrix.org/ says my server is fine, and it's listening on that port.As there's no error messages on starting PostgreSQL and matrix-synapse I don't really know where to go from here. Is my server just fucked? Do I have to start again from scratch? Or is there something I can do to save it? Please help.And yes I know I should stop being a cheap ass and just have a backup to prevent this from happening again.
