FMS on macOS - backup won't purge 'purgeable' data

Questions
,
1

I have a mac Mini running macOS 15.7.3 and FMS 22.0.4. All is well, except occasionally I'm getting warnings due to backup failures, such as this:

FileMaker Server 22.0.4 on example.com reported the following event:

2026-01-10 01:15:02.458 -0800	Error	160	example.com	Schedule "Weekly" was aborted; not enough free space on destination.

And indeed, in the FMS Admin webpage, the disk looks nearly full:

image
image506×158 27.6 KB

image
image724×734 31.4 KB

My database is about 15GB in size, so I could see how only having 34GB free space could cause a backup failure.

However

In the Finder, the disk shows over 100GB free:

image
image564×152 58.2 KB

Disk Utility shows more detail:

image
image758×182 39.3 KB

So the "free" space of 102GB is somewhat of a lie, as it contains 68GB of purgeable data.

My question:
Shouldn't FileMaker Server, when attempting to do a backup, be able to trigger the purging of this "purgeable" data? If not, why?

2

While it seems fair that FMS should be able to say, "I need that space", making backups is low priority. You're asking for a setting that says, "FMS backups is more important that you think."

And in a situation where capacity is running low there may be other ways to handle this. Is the backup trying to run at a time the machine is busy with other things? Is it easier to reduce the load on the machine? To reschedule? To provide more resources, RAM or disk space?

3

It's not clear to me whether:

  • FMS asks for the amount of free disk space, learns it is too low (because it's not accounting for purgeable data), therefore doesn't even attempt the backup (so the OS never even knows that it's time to purge data) vs.
  • FMS tries the backup, the OS doesn't purge data, backup fails.

Yesterday, I used Disk Utility which showed that the purgeable data was entirely APFS Snapshots, of the format com.apple.TimeMachine.2026-01-10-123456.local

I deleted these all, and purgeable space went to zero.

Today I look this morning, and it's filled up again - the drive is showing only 32GB free, with over 200GB purgeable:

image
image762×180 39.7 KB

The culprit again is time machine snapshots:

image
image1112×318 61.6 KB

Each hourly snapshot is about 8 to 15GB larger than the last one, which shouldn't be happening, since I excluded all the FileMaker data folders from TimeMachine. Perhaps I missed a folder? Will investigate.

4

Sounds strange, that your snapshots are this big.

Do you have a Time Machine disk connected for backups? Usually the local snapshot is deleted when it got copied over to the backup disk.

Otherwise you may write a shell script to run regularly on your Mac and delete the oldest snapshot automatically to keep e.g. maximum the last 3.

5

More info

  • Yes, my Time Machine backup drive is constantly connected. Like @MonkeybreadSoftware, I think it is strange that it's keeping over a dozen local snapshots.
  • When I check the actual backups being written to the TimeMachine drive, it looks as if my exclusion list is being followed, and the backup delta is under 1GB. The tmutil compare command compares the current drive to the last backup, showing what is going to be actually written:
sudo tmutil compare
[... many lines ...]
-------------------------------------
Added:         31.9M
Removed:       412.7M
Changed:       476.0M
  • however, I have seen theories that for local snapshots the exclusion list is ignored. If true, this would account for my local snapshots being 8G to 15G larger each hour (since my active FileMaker databases are about that size).
  • you can script cleanup using the thinlocalsnapshots command, with a Size and Urgency parameter (Urgency level 4 seems to mean 'very urgent')
tmutil thinlocalsnapshots  / 100G 4
Thinned local snapshots:
com.apple.TimeMachine.2026-01-11-174325.local

All in all this is rather mysterious. I'm still not sure whether this is a macOS bug (purgeable space is not being purged when FMS asks for it) or a FMS bug (FMS is asking for the wrong 'free space' value, and erroring out prematurely)?