Time Machine Goes on a Feeding Frenzy
My Mac’s Time Machine backup disk got full. This shouldn’t be a problem, it’s only 2TB and can’t keep everything forever. So it did it’s job and dutifully deleted the oldest backup to make room for the new one. Except it was still too full, so it deleted more backups. It got to the stage of only having a single backup snapshot left. It at least realised that it shouldn’t delete the last one, as it would stop being a backup at that point. Yet it was still completely full. Somehow a single backup of my 800GB internal drive data was taking up 1.9TB. Every hour a notification popped up telling me the last backup failed because the drive is full.
I glance with Finder to see what’s on the disk, only one backup from 4 days ago is there. It’s using 589GB on the disk. Time Machine only listed the one backup. sudo diskutil apfs listSnapshots disk6s2 only showed the one. tmutil listbackups only showed the one. BackupLoupe only showed the one.
And yet my disk was full. Finally while checking with different tools I found a bunch of .interrupted folders taking up a huge amount of space. These are only visible from the Terminal, checking the sizes via du suggests these must be where all that space has gone!
1.7T /Volumes/Time Machine 4.0G /Volumes/Time Machine/.Spotlight-V100 538G /Volumes/Time Machine/2026-01-17-142249.previous 429M /Volumes/Time Machine/2026-01-17-151714.interrupted 524G /Volumes/Time Machine/2026-01-17-151725.interrupted 538G /Volumes/Time Machine/2026-01-17-165608.interrupted 180G /Volumes/Time Machine/2026-01-17-200808.interrupted 2.0M /Volumes/Time Machine/2026-01-17-205341.inprogress 4.0K /Volumes/Time Machine/backup_manifest.plist
Not wanting to mess anything up I tried deleting these interrupted backups with tmutil, surely the proper way:
sudo tmutil delete -d '/Volumes/Time Machine' -t 2026-01-17-151714 (null): No such file or directory (error 2) POSIXError(_nsError: Error Domain=NSPOSIXErrorDomain Code=2 "No such file or directory")
It didn’t know what I was taking about. Okay fine time for the big guns.
sudo rm -rf /Volumes/Time\ Machine/*interrupted sudo rm -rf /Volumes/Time\ Machine/2026-01-17-205341.inprogress
For good measure I deleted a bunch of stuff on my internal drive that I didn’t need any more. You know, LLM model files, unused apps, stolen data dumps, the like. I then 'thinned' the local snapshots so that all the stuff I just purposefully deleted doesn’t clog up the next backup. This took my internal drive from ~800GB to ~400GB usage.
sudo tmutil thinlocalsnapshots / 1000000000000 4
The big number is the purge_amount, I wanted everything gone so I put 1TiB. The 4 is the urgency (1-4). It is unknown outside of Apple if 1 or 4 is the highest urgency, I ended up repeating the command with growing numbers for purge_amount and flipping urgency until tmutil listlocalsnapshots / showed nothing.
All sorted, I triggered a backup and it happily completed.
So What Went Wrong?⌗
This is actually the second time this has happened, which is why I spent a moment trying to find the problem. Unfortunately MacOS only keeps a day of logs, and thinning the local snapshots took away any chance of recovering what happened 4 days ago when all those backups got interrupted 🙄. Maybe I pulled the drive out while a backup was happening and that triggered the problem. Though I’m pretty careful about always unmounting first.[1] I’ve even seen it wait for time machine to cleanly finish canceling a backup before unmounting a few times.
Going from the .interrupted folders left behind, I’m worried that Time Machine just shat the bed on it’s own. It seemingly started a backup, copied 429MB, and 11 seconds later started a new backup. Instead of resuming the incomplete backup, it started a new one, while leaving the incomplete one.
It then copied over 500GB of data, somehow, despite the entire internal drive having 800GB of data on it. Surely I didn’t create 500GB of new data within 1 hour. It then went on a loop doing this until the drive hit its 1.9TB quota.
I’m guessing that it gets into this knot when running out of space during a backup, so it stops and deletes old snapshots. Then it starts again but runs out of space again because it didn’t clear the failed backup. It’s very fustrating since the switch from HFS+ to APFS and using snapshots changed how time machine and tmutil works slightly, making most of the online advice not applicable. It’s also annoying since much of that advice boils down to "Yeah it eventually just craps out, just format the backup drive and start with a clean slate".