When I run Disk Utility First Aid on Macintosh HD – Data (of the internal boot drive), every snapshot is followed by a series of warnings, e.g.,
Checking snapshot 7 of 13 (com.apple.TimeMachine.2025-08-06-192011.local, transaction ID 11247853)
warning: physical_size (352256) of dir-stats object (id 1051246) is greater than expected (110592)
warning: descendants (5213) of dir-stats object (id 24687880) is greater than expected (5209)
etc…
At the end of First Aid, Disk Utility reports “repaired,” but the warnings remain after another run of DU. I have not tried repairing from Recovery Mode or from the Container because every new snapshot receives the same warnings. Thus, the OS continually creates them.
Of course, I don’t have any idea about the severity of the situation.
Apple Discussions suggest that the issue first began with Sonoma (I updated to Sequoia directly from Ventura). Seems like by now someone would have reported this issue to Apple. Before I do so, I wondered whether TidBITS readers might have some insight.
If the warnings are definitely only referring to the snapshots you could try deleting all snapshots and see if they reoccur on new ones. In my experience they often don’t, but may do.
“In macOS Ventura or later, click Options, choose Manually from the Backup Frequency pop-up menu, then click Done. Wait a few minutes to allow local snapshots to be deleted.”
I waited 30 min, but the snapshots (and warnings) remained.
Also, First Aid always ends with this:
“The volume /dev/rdisk1s2 with UUID 85… was found to be corrupt and needs to be repaired.
Verifying allocated space.
Performing deferred repairs.
The volume /dev/rdisk1s2 with UUID 85… appears to be OK.”
If “repair” refers to the TM snapshots, DU is hallucinating.
Thanks for your feedback. Perhaps I’ll try repairing in Recovery mode.
Do that. Also make sure you do First Aid three times: 1) the drive, 2) the APFS container/partition, and 3) on the data volume(s). (From your first post you seem to have only done First Aid on the data volume.)
Note that these instructions have changed recently. Before we learned that we should run it top down. But as of Sequoia, you should start with the volume, then go to the container, and finally run it on the drive.
Note that the Apple article that @Simon notes says that running DFA on the startup volume or disk should be done from Recovery. I believe that to be a requirement, not a suggestion. I’ve seen DFA show errors on the startup disk when running “on-line” (i.e. when running on a booted macOS) that don’t appear when run from Recovery.
Also consider that if DFA finds a problem, you don’t want it to make repairs of a file system at the file system structural level when that file system is mounted and running. That’s a corruption potential waiting to happen.
Thanks to everyone for the feedback. I’ll implement the instructions when I have some free time. I have only the one device, and I’m in the midst of a major research project (aren’t all projects major). Just wanted to say thanks now.
Ran DFA on the Data volume in Recovery Mode. It showed all the aforesaid warnings, errors, etc. for the local TM snapshots. It claimed to have repaired them (but I knew better).
Then ran DFA on the Container. No errors reported.
Then I tried to run DFA on the drive. A slew of errors and warnings appeared. I believe they included “Quick! Make a backup” and “Could not unmount the drive.” Well, at that point I backed out and re-booted normally.
I then ran DFA on the Data volume under normal boot. All of the warnings, etc. for the local snapshots appeared. I was not surprised.
However, Apple says “If at first you don’t succeed, try one more time.” OK, back to Recovery Mode. Amazingly, the local TM snapshots were error-free! Well, that was enough for me. Although I probably should have done so, I didn’t bother with unleashing DFA at the Container or Drive level. Maybe another day.
Finally, I ran DFA on the Data volume under normal boot. All the snapshot warnings and errors were gone, thankfully.
I don’t understand any of this. But many thanks for all the advice.
I don’t know the solution to your problem, but I can add a little:
This is probably because the Recovery partition you booted from is also on that drive. And yes, you shouldn’t repair anything on a volume you can’t unmount (DFA might not let you).
I wonder if something just blew them away.
Local snapshots are created hourly by Time Machine. The system only retains the most recent 24 (one day’s worth), with the old ones getting deleted to make room for the new ones.
If there are corrupt snapshots, then deleting them would “solve” the problem. And since local snapshots should all go away after 24 hours, I would expect this problem to solve itself.
Maybe doing a repair from Recovery mode deleted those snapshots? That’s the only thing I can think of.
Thank you David. When I first encountered the issue, I had a mix of snapshots from the prior day plus the day of. All of them produced the same warnings/errors. The same was true last night when I did the Recovery Mode repair. So it seemed that the OS was continually creating wonky local snapshots. That behavior seems to have stopped (for now?). There are two additional snapshots, and all are reported OK.
My mini doesn’t travel, and it’s always tethered to two backup drives, TM and CCC, which cover far more territory than 24 hours. So having valid local snapshots is not a priority. I was just concerned about there being several Gb of corrupt data on the boot drive, even if I never attempted to use that data.