Open Bumpay opened 4 months ago
It's implemented using rsync --link-dest
such that it does incremental backups but utilizes hard links to give the behavior of each backup being full contained.
Ah alright, I was looking at the directories through FileZilla and thought, they are all backing up redundant parts of the world.
But running du -sh *
shows, that the backups are actually really small in size. Thanks for pointing that out.
Here is the result of the du -sh *
command:
16G world-20240703-080010
3.8M world-20240703-080101
3.8M world-20240703-081010
3.8M world-20240703-082010
3.8M world-20240703-082418
3.8M world-20240703-091909
81M world-20240703-111410
237M world-20240703-130911
105M world-20240703-140912
106M world-20240703-150914
142M world-20240703-160915
435M world-20240703-170916
By the way, how does it prune the backups with the rsync backup method?
It uses a find
by modified time and does rm -r
on those directories
Being hard links it works out to prune away each top level backup directory.
Ah okay thanks, so to confirm I understand that correct. As long as there is a backup that hard links to the top level backup the top level backup will not be pruned?
Then, when are top level backups created? Is it just once in the beginning? And what would happen to the linked backups if the top level backup is being removed?
Sorry if I don't understand the (hard) linking, I'm fairly new to working with Linux.
Understandable, hard links don't get talked about much but they're a pretty powerful feature in this scenarios. This description is pretty good but was hoping to find something that explained more concisely
As far as I know rsync should only do one full backup and then have only backups of the changes. But why do I get only full backups all the time? Did I configure something wrong?
compose.yml