/archive/maps/staff-notes/datapacks/datapacks-plugins/ · last edited 19 Apr 2020 · Development
The primary concern is the stability of the world save files since we are running on a shared hosting plan with limited disk space. We agreed that a full dump of the world folder must occur every six hours rather than daily to minimize data loss in case of corruption or griefing. The current script that runs automatically is unreliable because it conflicts with the server save command, so we need to stop the server before copying the files. This means we will implement a manual trigger via the console for now, which requires one moderator to be online during the backup window.
We discussed the storage location for these backups. Keeping them on the same physical drive as the live server is too risky, so we decided to copy the compressed archives to a separate cloud storage bucket. The process involves creating a zip archive of the world folder, uploading it to the cloud, and then deleting the oldest archive if the storage quota is exceeded. This rotation should keep us within our monthly limits while retaining at least three days of history.
There are open questions regarding the automation of this process. We need to determine if a simple cron job can handle the stop-copy-start sequence without causing excessive downtime for players. Another issue is the naming convention for the files. We want to include the date and time in the filename to make restoration easier, but we also need a way to mark specific backups that coincide with major updates or events.
The team needs to test the restoration process in a test environment before relying on it for production. We must verify that the world loads correctly after being extracted from the cloud storage. If the test fails, we may need to adjust the compression settings or the file permissions to ensure compatibility. Until this is confirmed, no one should attempt to restore a backup in the live server without explicit permission from the head moderator.