57 comments

[ 2.6 ms ] story [ 53.3 ms ] thread
I use the same setup and was able to restore some files I recently deleted. My SMB settings in Synology were set to what the recommended settings were already. Not sure what happened in this person's case, but it also seems like he backed up and didn't test the restores. Which isn't good practice.
macOS yearly updates haven't been great since they started but Tahoe is a new low.

Apple really needs to turn things around.

Another disturbing example of sloppy execution by Apple Software Engineering. This only reinforces my resolve to avoid upgrading to macOS Tahoe.
Apple has broken Time Machine enough times that I would never consider using it at all anymore. Once upon a time, it was really neat, had great integration with Mac OS X, and an amazing user interface and experience, but it's now clearly technology that Apple will probably eventually drop entirely in favor of something less impressive all together, like telling you to buy more iCloud Storage.
If you set your Apple device to beta updates for the previous release you can suppress the constant prompts to upgrade. Reduces the chance of accidentally upgrading.
Article title is a bit dramatic. The summary seems to be: for the 5% of users who back-up to a network share (rather than direct-attached storage like a USB hard drive enclosure), Apple's default SMB configs on Tahoe are strict and won't work out of the box with many common NAS solutions.

Apple should document such changes, but, looking at the post title, you'd think they were silently corrupting data during restoration.

The author posted a fix, but how do I check if there is a problem in the first place?
Time Machine has always been a bit ropey on SMB shares. I think it’s in part because it creates a disk image on the share then writes to that. This creates a lot more work and potential for things to go wrong.

If you want to backup across the network then it’s probably best to choose some third party software.

I have been trying to trouble shoot a Time Machine issue since upgrading to Tahoe. It is usb backup. So far none of the most recent stated fixes work.

An initial backup on newly formatted disk will run but very slowly. Perhaps reaching 100% but it never finishes. At some point the percentage will change and the backup will stay stuck at somewhere near 10%. Cancel backup and run it again. Gets to ~10% and stays stuck. Multiple drives. Re-fs'ed. Boot into safe mode. Networking off. Etc, etc. etc. The TimeMachineMechanic app doesn't have any revealing feedback. I can run a full tar backup to the same disks.

No idea.

I haven't tried backing up to a network share but really, it shouldn't be this difficult.

Clearly someone didn't test a bunch of edge cases when pushing this one out.

In a terminal window run

  log stream --predicate 'subsystem == "com.apple.TimeMachine" AND NOT (category == "LogLimits" OR category == "VolumeViewModel")' --info --debug --style compact
and then start a backup (either from the menu bar icon, the system settings panel, or "tmutil startbackup"). This will tell you what Time Machine is doing, and might give you some useful information.

  man log
where you can use "show" and a lookback period instead of "stream".

  man tmutil
is pretty decent documentation, although the glossary secdtion ("BACKUP STRUCTURE") is important to understand if reading the whole man page.

Some things to look out for are what filesystem your newly formatted external volume is (APFS might not be great for a single spinny disk, for example), and what version of USB is in use (friends don't let friends do USB 2 mass storage). With inexpensive external media it's often a cable or power supply issue, even if (as in your case) tar appears to work. Have you checked that the contents of the tar file are correct? Also, tar files tend to be streamed out to sequential LBAs, where smaller files and (in Time Machine backups) holes might lead to a different write pattern that the drive might not like. Maybe test with rsync -c instead of tar?

The bigger question is why does Time Machine continue use a network file system for backups? It's so fragile you can't rely on it. It's gotten better in recent years, possibly due to APFS, but that just means somewhat longer intervals between disasters (wipe out and reinitialize, losing all your backups). A T.M. using a custom protocol to save and restore blocks would fail sometimes too, but not ruin all your existing backups.

edit: I use Arq for daily backups, but T.M. for hourly. When T.M. eventually craters its storage, I have robust dailies in the cloud, so no worries.

Outsider perspective here (never used Time Machine), but my first thought is that rsync works amazingly both local and over the network. Can't imagine why it being over the network would be a problem. If it can resume a partial transfer and compare checksums to ensure a match, what's the problem?
It's backups, not sync, so it needs to support things like incrementals/differentials, deduplication, retention rules, thinning, etc.
I'm a big fan of SuperDuper [1]. I use it for daily differential backups to a secondary SSD. I don't get the hourly backups that TimeMachine has, but my SuperDuper backups are directly bootable in the event that my system disk dies.

I'm sure you could do the same with cron and rsync, but I can't be bothered.

[1] https://shirt-pocket.com/SuperDuper/SuperDuperDescription.ht...

Time Machine is held in high regard for some reason (maybe the fancy scrolling interface when you look for files to restore?) but it's not really useable. It pretends that backups-over-the-network are a possibility but its completely unstable over the network and invariably decides the backup is corrupt after a few months and then tells you you have to start from scratch.
With external SSDs plugged directly into a USB port, it's worked 100% fine for me and saved my butt a few times.

But, I haven't installed Tahoe. I may skip it entirely, hoping that they do a Snow Leopard-like clean-up-the-mess release in September.

I finally got fed up with TM and switched to borg via Vorta. So much more reliable. A couple of times I've gotten error messages when I went off network while it was trying to do a backup, but each time the repo was fine.
If you set Time Machine to use encrypted backups, it will create a fake disk that's really a directory tree with a bunch of gigabyte-sized binary chunks. This is safer because it doesn't require the file system to support anything fancy like symlinks or case-insensitive unicode file names. One downside is that restoring to anything other than a Mac is nontrivial.
Unencrypted sparse bundles for TM can also be created for non-TM supported network locations.
It reliably kernel panics since tahoe at a certain point
restic and kopia should work decently, if with a bit of setup, I think both can just mount backup as FUSE filesystem

The backup system that silently breaks when it doesn't like something in backend is not worth time

On Linux we get this for free with Btrfs copy on write snapshots, snapper, and Btrfs Assistant.
Apple has always had problems with SMB since they switched from one of the open-source implementations to one it internally developed, many yaers ago.

Then again, SMB especially in its newer versions seems to be a protocol developed by MS with one of its goals being to make third-party implementations as difficult as possible.

I have the same setup and it works fine on my machine. ¯\_(ツ)_/¯
I had so many corrupted Time Machine backups over the years that I eventually just wrote an incremental backup script in rsync. I’m much happier.

Something like [1] can be inspiration.

[1]: https://github.com/perfacilis/backup

I had the exact same experience and did the exact same thing. I also moved my backups from HFS+ to zfs, and got more serious about my backup strategy.

If I had to start over I'd go with rustic-rs or borg backup.

FWIW I do still use `tmutil localsnapshot` for local macOS snapshots where you can use the Time Machine UI to restore files.

Just for the record: I wanted to see your content, but I couldn't because in Spain when there's football they block most websites to "avoid illegal football IP lists"... LaLiga can block anything they want without any restriction, even you website which I doubt about it. I can barely navigate... I will read it later tomorrow. This why you might see 0 traffic from Spain.
On Tahoe my Time Machine was broken after the update. My backup target is on a QNAP NAS. I just had to set it up from scratch again and it worked. But it did cost me a few files I was trying to recover. So I feel this.
As someone from Tahoe, it makes me sad that the release with the worst reputation is named after my home region.
I never trusted Time Machine, my primary line of defense is rsync to a server running ZFS with hourly snapshots, and weekly rotations of offsite drives. For bootable backups, Carbon Copy Cloner.
+1 for TM to ZFS - although over samba. A ZFS snapshot at each successful disconnection means any occasional corruption is simple to rollback. Also I’m using nightly Arc backup to B2 of critical files.
Tahoe backups to UnRAID's native Time Machine backup system (as described at <https://docs.unraid.net/unraid-os/using-unraid-to/manage-sto...>) does not work in UnRAID 7.2.3. It is not (solely) caused by Tahoe, however, because it did work in 7.1.4. <https://forums.unraid.net/topic/195091-time-machine-backup-d...>

mbentley's Docker image version of Time Machine—which I began using back when native Time Machine support was completely broken <https://www.reddit.com/r/unRAID/comments/16x3ddm/my_experien...>—which the post mentions is unaffected, and continues to work with Tahoe without configuration changes.

What is Apple’s QA process? Do they rely on some random set of manual tests that may or may not get run each release? There have been so many things that seem like one of the most valuable companies in the world would include in tests, but yet break or remain broken.

As an experiment, open Console and filter just errors and faults. Dozens to hundreds of “errors” will scroll by representing the normal operation of the system. (Either they’re not really errors and no one cares or they really are errors and Apple just leaves their systems broken). How can anyone think this is OK?

I haven’t upgraded to Tahoe. I have been a Mac power user for over 20 years, and it becomes less interesting every release. I came for Unix, the script ability, and 3ᴿᴰ party applications. Unix is an afterthought, script ability is all gated behind security gates, and modern apps seem like such a huge regression.

> What is Apple’s QA process?

"Does this increase iCloud subscriptions or not?"

They didn’t have qa for MacOS when I might have known someone who could speak on that. Was a shocker then, but no surprise now.
(comment deleted)
A colleague joined the team for one of the most visible features in MacOS a few years ago and told me they had no automated testing.
Most macOS teams have unit testing. The quality of which varies greatly.
I think they mostly test in an all-apple environment.

With third party stuff, maybe you'll get lucky, but no guarantees...

3rd party monitors, or keyboards, or mice (what's a mouse?) or ...SMB devices