Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

A superpower of doing Time Machine backups to a Synology is that you can put it on a BTRFS drive with automatic snapshotting. About once a year something seems to happen that causes Time Machine to give a “whoopsie, gotta start over from the beginning again!” sort of error. I’ve found that going back a day or two with snapshots on BTRFS fixes the problem.


For me it was more like every day. I gave up on Time Machine very quickly.

The underlying problem is that Time Machine stores a disk image over the network[0]. If the disk image is not unmounted cleanly, Time Machine treats it as horribly corrupted and refuses to touch it. And this happens very often if you're using a laptop that will be unexpectedly disconnected from its storage constantly.

[0] This is to support things like hardlinks, etc. Ironically this is to emulate what BTRFS snapshots do natively. No clue if modern Time Machine uses APFS, but so long as they shove the actual data inside of a disk image this problem will continue occurring.


I've given up on it and bought a SSD to do local backups. My new plan for the Synology is to get Minio running on it and then use Arq Backup to send the data.


Update: This is working quite well. I occasionally run into errors "Error: The network connection was lost." but I think this is mostly because of my Wifi.


> About once a year something seems to happen that causes Time Machine to give a “whoopsie, gotta start over from the beginning again!” sort of error.

This is true. Though I've had very good luck re: ZFS and my current Samba config.

If you're not already, you should be snapshotting before and after each network mount. See: https://kimono-koans.github.io/opinionated-guide/#on-network...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: