This feature is really exciting! Looks like it’s gonna save us a lot of hassle. I’ve got backups working for all of my staging clusters, but I decided test restoring from a backup before enabling this for my production clusters, and that’s when I ran into an issue. For one of my clusters, when I press the “Restore DB from backup” button it creates the new app as expected, but the new machine never gets to a healthy state. In the live logs I’m seeing things like
[info] restore | ERROR: Connection problem with ssh
and
[info] restore | 2024-08-28 22:53:49.737 UTC [359] LOG: invalid checkpoint record
[info] restore | 2024-08-28 22:53:49.737 UTC [359] FATAL: could not locate required checkpoint record
[info] restore | 2024-08-28 22:53:49.737 UTC [359] HINT: If you are restoring from a backup, touch "/data/postgresql/recovery.signal" and add required recovery options.
[info] restore | If you are not restoring from a backup, try removing the file "/data/postgresql/backup_label".
[info] restore | Be careful: removing "/data/postgresql/backup_label" will result in a corrupt cluster if restoring from a backup.
and eventually the machine gets into a state where it just keeps logging things like
[info] failed post-init: failed to establish connection to local node...
Any pointers on what the issue could be would be greatly appreciated.