1

I have a new MacBook Air running 10.14.5 (Mojave, non-beta). The drive is 500GB and formatted APFS.

The initial Time Machine backup has been running for about 24 hours and still isn't finished.

But it has backed up more data than could ever possibly exist on the drive:

enter image description here

Note that it says it's backing up 683 GB (and increasing) of a 500 GB drive, when not even 200 GB of the drive has been used.

I checked the sizes of my local snapshots, and there are only 4 of them, with the following sizes:

  • 502.2 MB
  • 217.5 MB
  • 5.4 MB
  • 7.4 MB

So that doesn't come anywhere near explaining the difference.

The Time Machine drive was cleanly formatted as HFS+ right before the Time Machine backup was initiated.

So… why is it using so much space? How is it using so much space?

TJ Luoma
  • 20,659
  • The incremental stores can use twice what you expect when things get interrupted. You may need to let this complete and attach a program like BackupLoupe to analyze the files intelligently. Or you could wipe the backup and set up exclusion and only let a small portion of your home folder (Music or Photos) to get a good baseline before backing up everything. – bmike Jul 06 '19 at 17:50
  • That's really weird. It's now up 1.22 TB and the drive is only 1.25 TB, so I'm not sure what's going to happen when it fills up on the first backup. I wonder if I should turn off Time Machine entirely and then turn it back on. – TJ Luoma Jul 06 '19 at 19:56
  • Did you ever solve this? Hopefully you didn’t need to follow a “nurse it along” process like described here. – bmike Oct 03 '20 at 20:17
  • There was no solution, really. I try not to interrupt Time Machine if at all possible. Since I have a MacBook Pro which I sometimes need to close and take with me, I have a Keyboard Maestro script which tries to unmount the Time Machine drive, starting around 6:00 a.m. so that it won't be running by the time I need to leave the office (back when that was a thing). – TJ Luoma Oct 04 '20 at 11:34
  • Thanks! I throw hardware at Time Machine for this very reason, you do play roulette each time a cable gets bumped and the more intervals you have, the longer the disk check takes when you reconnect them. Having several destinations in rotation let’s me not lose everything when a backup disk fails or just takes too long to be “checked”. – bmike Oct 04 '20 at 12:41

1 Answers1

1

Not exactly sure since I haven’t used TM with Mojave yet however, I may help point you in the right direction or stimulate a more accurate answer. TM can backup all files including temp files, deleted files, cache files and even USB drives attached. If you were making a lot of changes during the initial backup, the initial file size will compensate for that. Also Mojave utilizes APFS which contains nested volumes and containers. There are additional formatting files outside of the readable HD within the volumes and containers that could be getting backed up to make sure you will be able to install from backup correctly. It could also be backing up your recovery HD. If you run diskUtil list in terminal you can see the size of any hidden volumes and containers.

I wonder since APFS uses the new file reference system to save space while keeping track of duplicated and edited files, maybe that could be play a role as well. TM HD might be saving both or be adding additional reference files for the locals snapshots? In the long run, TM will be writing over the older saved backup files with newer content. So I don’t think you should pry too much to get that additional space back.

I have had issues in the past trying to restore from a TM HD that was the same size as my primary HD and wouldn’t recommend it. Using a drive at least 2/3 times larger is the right way to go.

Theologin
  • 530
  • 2
  • 7
  • I’d remove the container speculation, I use TM all the time on Mojave and the source containers are reliably solid in general. Unless TJ has a complicated disk utility list the other avenues might be more fruitful to chase down. Great ideas and it still could be APFS corruption, just unlikely. – bmike Jul 06 '19 at 17:52