5

I have a new iMac running OSX 10.12.4. It has an internal Apple-Supplied 512Gb SSD, of which about 140 Gb is used. As with my previous machine, I use Time Machine to backup on to an external 2Tb hard disk.

However, I find that every TM backup appears to be a full backup, not an incremental backup.

How can I get Time Machine to make proper incremental backups, as it’s supposed to?

By way of background,

  • Each backup takes a very long time; so long that it cannot backup hourly because it’s still completing the last backup.
  • On the Preferences panel it tells me that it is backing up … GB of 131 GB.
  • The space left on the drive corresponds to multiplying the number of backups by the full amount of data.
  • After about 18 backups, it starts deleting old backups
  • I get the disk full message too soon

So I am convinced that they’re all full backups.

I have tried everything I can think of, including:

  • turning it off and on again
  • re-formatting the drive
  • replacing the drive
  • reinstalling the OS
  • deleting the plist, as per Pondini’s web site
  • removing some software suspected of interfering with TM
  • excluding additional folders
  • putting chicken feathers in a paper bag and waving it around my head

I have spent countless hour with Apple support. As a desperation move, they now want me to do a clean install, which, in my opinion is going too far, and certainly not something I am enthusiastic.

I am aware of this question: Time Machine keeps doing full backups

but

  • it’s 5 years old
  • the asker says that TM also does incremental backups
  • despite the accepted answer, the question is not really answered
8
  • What does BackupLoupe say about the last backup which was made? (Seems like this is the same suggestion in the linked post you added in grace period! I'd still be interested to see what it says.)
    – grg
    Commented May 13, 2017 at 11:44
  • @grgarside Hard to say. It takes a very long time to calculate sizes, which is not a good sign. When it does, each backup is the same size as the original.
    – Manngo
    Commented May 13, 2017 at 11:46
  • Have you tried analysing the last backup using the load button next to the backup in the sidebar (or is that what you meant by "a long time to calculate sizes")?
    – grg
    Commented May 13, 2017 at 11:47
  • I have just opened it again, and asked it to re-index everything. However, yes, that is what I mean.
    – Manngo
    Commented May 13, 2017 at 11:50
  • @grgarside BackupLoupe has finally given me the results on the first 4 backups: all about 130 Gb. Meanwhile there are some more being queued.
    – Manngo
    Commented May 13, 2017 at 23:32

4 Answers 4

1

Backups are managed via Apple File System Events (not Spotlight) https://en.wikipedia.org/wiki/FSEvents

It seems most likely that something is touching your files in a way that convinces TimeMachine a new copy is needed. There could also be something damaging the FSEvent cache which would cause TM to create a new backup.

You might want to check the system console log and see if there are any messages there that hint at the underlying problem.

2
  • Thanks for your comment, giving me somewhere else to look. Would that be related to the Not using file event preflight message I’m getting in my logs?
    – Manngo
    Commented May 26, 2017 at 5:28
  • I haven't seen that message but it sure sounds plausible from the wording. See what Google says Commented May 26, 2017 at 14:03
1

Possibly related and possibly helpful to others?

I've been fighting Time Machine only doing full backups on my M1 Mac mini for 15 days now and have finally prevailed. Mac mini with one external drive for extra storage and Time Machine backups.

I have three 8 TB Time Capsules, one in a shed in the back of the garden. 15 days ago I noticed that Time Machine suddenly wanted to backup 700 GB. The next backup again would backup 700 GB. And so on.

This is what was present in all the Time Machine log, every time:

 TimeMach    Couldn't find event marker for volume [UUID]. Volume store not present in [MACHINE DIRECTORY]
 TimeMach    First backup of source: "Data" (device: /dev/disk3s5 mount: '/System/Volumes/Data' fsUUID: [UUID] eventDBUUID: [UUID]) firstBackupOfVolume

It happened on all backup devices. Creating a new backup didn't help. Safe-boot, re-installing macOS and resetting all Time Machine preferences didn't help either. The problem occurred on all back-up devices. Even on a new external HDD I used as a new backup destination for testing.

Logfiles also showed the external HDD did do incremental backups, only the internal Data volume of my Mac mini was problematic.

I took quite some time crawling through logs, looking using fs_usage, and examining the new way backups are made. Yesterday I suddenly noticed Extended Attributes on /System/Volumes/Data that were suspect:

 com.apple.backupd.BackupMachineAddress: [MAC ADDRESS]
 com.apple.backupd.ComputerName: [COMPUTER NAME]
 com.apple.backupd.HostUUID: [UUID]
 com.apple.backupd.ModelID: Macmini9,1
 com.apple.timemachine.private.structure.metadata: APFSMachineStore,1,[UUID]
 purgeable-drecs-fixed:

These com.apple Extended Attributes were not present on /System/Volumes/Data of any other Mac I checked. Similar attributes are present however on the Data directory of the backup.

After deleting all suspect attributes (except for "purgeable-drecs-fixed") and deleting all intermediate en interrupted backups, my backups are finally incremental again.

 xattr -d /System/Volumes/Data/com.apple.backupd.BackupMachineAddress
 xattr -d /System/Volumes/Data/com.apple.backupd.ComputerName
 xattr -d /System/Volumes/Data/com.apple.backupd.HostUUID
 xattr -d /System/Volumes/Data/com.apple.backupd.ModelID
 xattr -d /System/Volumes/Data/com.apple.timemachine.private.structure.metadata

I suspect these surplus attributed were copied to the Data directory of the backup data structure during the backup process, overwriting attributes used by Time Machine, and Time Machine would not recognise the previous backups as belonging to the same data during the next backup.

The big question is how did these Extended Attributes end up in the wrong place? The day the problems started I had updated from macOS 14.5 to macOS 14.7. I had also played with Docker running PiHole. The latter has an issue, a conflict with mDNSResponder listening to 53 instead of 5353 when PiHole starts. For a test I had temporarily unloaded the mDNSResponder launch item. This caused havoc with iCloud Documents and probably other services. Maybe this caused backupd to lose track of the correct Data directory to apply Extended Attributes to?

0

My gut reaction is that this sounds like Spotlight isn't indexing properly. Spotlight is how the system knows which files have been backed up and which files have since changed or been added. My first step might be to reset Spotlight by opening Terminal and running this command:

$ sudo mdutil -E /

that'll delete /.Spotlight-V100 and rebuild it on your main volume. Failing that, head on over to http://pondini.org/TM/D7.html - that website is fantastic for troubleshooting Time Machine issues, and that page is dedicated specifically to the problem you're experiencing.

2
  • @Manngo If you read through and do the steps and find that something there fixes it, post the answer back here and mark it as the correct answer so that people in the future have a good chance of trying your steps first.
    – Harv
    Commented May 17, 2017 at 7:26
  • Thanks for the suggestion. I think I’ve tried everything on the pondini site, but I’ll have another look. Meanwhile, I have reset Spotlight as you suggested. Of course it will take a while to see whether it works, so I’ll let you know how it goes.
    – Manngo
    Commented May 17, 2017 at 8:08
0

This doesn’t resolve why the problem arose in the first place, but it certainly fixed the problem.

After spending many hours with Apple support, somebody finally suggested that I reinstall the whole thing and start again. Needless to say, I was reluctant to try that.

However, an Apple Genius whom I consulted on a different matter did mention that doing this and restoring from a Time Machine backup to a fresh setup actually does things differently.

I had enough space on my internal disk to partition it so I could try it out without losing everything.

Here are the steps I followed:

  • Partition the internal drive using Disk Utility
  • Reboot into Recovery Mode
  • Install MacOS into the new partition
  • Reboot into the new partion and follow the usual process
  • When the time comes choose to restore from a Time Machine backup

There were a few minor issues with reconfiguring some things which TM didn’t have, but over all it worked smoothly. It is a pleasure to see it all working properly again.

Phew.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .