To Trust APFS or not

frederico wrote: “To clarify a bit further, am I incorrect that DragThing is still working under Mojave? No need to stall at Sierra, correct?”

Yes, I expect it does.

I’m distrustful of APFS in its current state. Since I have internal SSDs on most of my macs now, it’s not easily possible to avoid an unwanted conversion from HFS+ during system updates. I use monolithic images extensively both at home and at work; that often fails past Sierra and is officially deprecated by Apple. I have yet to see any disk repair utilities that support APFS beyond copying the same files that Finder would be able to copy, file space reporting is highly inconsistent depending on how you ask (Finder, command line, etc), it’s still not well documented…

I also haven’t seen any for-me compelling, or even interesting, features in the newer systems. I avoid clouds and sync, don’t want deep integration between apps or devices, and Mojave’s dark mode has crept so much into the light mode that I really can’t stand to look at it for more than a few minutes at a time.

I carefully bought a 2017 iMac right before WWDC to be sure of being able to be a stick in the mud, though I’d have preferred to see what the newer hardware would be like.

/end old fogey mode

For the Trash, I try not to let mine build up too much, though I’m not adept at housekeeping, either in real life or virtually. But I once accidently emptied a users trash, and they were not happy…

LOL! Remember when Norton made a fortune just by providing a Trash-undo feature? I think TechTool or Drive Genius or one of those still offers it.

I was wary of APFS at first, as well, especially when they yanked it back from HDD and externals and Time Machine volumes near the end of the beta cycle; seemed like a bad sign; however, as I was force-moved onto it, I find it exceptionally stable, and I don’t miss the random, regular breakdowns of HFS+. Further, when performing recovery of an otherwise “irreparable volume” as a result of incessant kernel panics (by users who should’ve known better) on more than one occasion I was able to recover data I feel pretty confident would not have fully survived HFS+ without a deep dive using Data Rescue and extensive time.

Besides, if your backups strategy includes both clones and another real time incremental system, it really doesn’t matter that much for most people. If APFS fails, you shouldn’t be in any more danger than if HFS+ fails, n’est-ce pas?

And APFS offers some truly incredible benefits, that I’m loving, personally. I’m willing to trade the space reporting issues and container confusion and occasional hoops and hurdles when reformatting becomes necessary to repurpose a drive.

And gosh knows all our iOS devices have been incredibly stable on APFS; when’s the last time you had to wipe an iPhone and do a system restore because of inexplicable slowdowns and random shutdowns? I’m giving a lot of credit to APFS.

I took a cursory look at using KM or BTT for emptying Trash on a given volume; easy peasy.

frederico wrote: “I don’t miss the random, regular breakdowns of HFS+.”

Ever since journalling, I’ve haven’t had much trouble with that. Pre-journalling was indeed a nightmare, though it certainly helped pay the rent :-).

In principle, I do approve of APFS, but it worries me that backup and other utilities have had to reverse engineer things instead of having actual documentation to work from. I’m sure it will eventually settle down. Meanwhile I have other software that will keep me on Sierra, notably Aperture and Filemaker 12.

“if your backups strategy includes both clones and another real time incremental system”

For me, I’m still working on it. My file server (data and system) is pretty well covered, but most of the system drives only back up to the time machine server (which is not backed up). I bought a couple of new drives last week to got offsite clones of the system drives. I’ve lagged because I try not to have anything of value on the system drives, and as far as getting things done until I have time to rebuild something, the odds of all of my systems dropping dead at the same time is unlikely–if it happens, I’ll probably have much worse problems to deal with first anyway (earthquake, house fire.)

As for users, I try to make them backup, and most use time machine at least some of the time, but I can’t force them. I’m not looking forward to the first T2 SSD that happens to fail…If only I could retire before it happens!

Just curious – are the T2s being used for Time Machine?


David Brostoff wrote: “Just curious – are the T2s being used for Time Machine?”

No, the T2 controls the internal SSD, doing hardware encryption, amongst other things. If there’s an up-to-date, uncorrupted TM backup, or a clone by other backup software, there’s no long term problem if the SSD eats itself–just get a replacement from apple if still under warranty, or buy a whole new computer if it isn’t (feh) and restore from the backup. But when it’s hard to get users to actually do the TM (or any) backups regularly , there could be no recourse to try to rescue any data. If you can’t mount the drive with target disk mode, it’s probably toast. Apple might be able recover sometimes, but I sure won’t be able to. With a soldered SSD and no T2, you can send the whole computer to Drive Savers or equivalent, and they’d at least have a chance. The T2 makes recovery much much harder–by intent. It keeps the bad guys out too, and that’s a good thing. But you must keep up to date backups, preferably several different kinds using different software, and that’s a hard sell to most non-computer people, especially if in the past it hasn’t been too hard to recover most data on a drive.

I’m hoping that it’s all quite robust, but it just hasn’t been around long enough to know about failure rates yet. Then there’s the issue that the fewer problems people hear about, the less likely they are to do backups at all. My only real converts are the ones who’ve lost a significant amount of stuff, and even they start to get careless again within about a year.

I had a couple users who got bit up to three times, and each time shed massive tears for the lost data due to their own laziness/neglect/lack of discipline.

That’s one nice thing about CCC, is that you can set it to report/email/prompt for “errors” when the backup volume isn’t present at the time of the next scheduled backup; usually those messages stack up enough for most people to get around to attaching the drive. (For family members, I CC myself on those errors, and nag them personally if I see a failure without a followup success in a timely manner).

But for Time Machine, it doesn’t provide any native warnings, of course, so for these two particular problem children, I wrote a script to monitor their TM backups, and if they didn’t complete one in a timely manner, I had it repeatedly and progressively post obnoxious, impossible to ignore warning messages such that the machine was eventually effectively unusable until they attached the drive or got on the network that hosted TM.

For users I actually am/was paid to manage, I have a “dashboard” on my Desktop (GeekTool) and via Web (HTML/ Dropbox) that monitors both CCC and TM and reports last-completed statuses/dates, and slowly changes from green to yellow to red for each so I have a quick visual to know when to chastise, or correct, if it’s a legit software, not user failure.

The irony of the T2 security, of course, is that 99% of people do not do anything resembling a secure, let alone encrypted backup, and that unencrypted backup drive is usually within hands-reach of the machine that might be stolen.

And I can’t tell you the number of people who give me a dumb blank look when I point out they are carrying their only backup in the same bag as their only laptop; encrypted or not, what’s your plan, Dexter, when someone snags your bag at Starbucks?

frederico wrote:

“And I can’t tell you the number of people who give me a dumb blank look when I point out they are carrying their only backup in the same bag as their only laptop; encrypted or not, what’s your plan, Dexter, when someone snags your bag at Starbucks?”

Yeah. My users are smart people. But they’re over-specialized and overworked, so it’s hard to get them to spare enough brain cells for what they consider to be modern pencils. And they’re right in principle–it’s the current state of practice that bites 'em.

I occasionally dream of having tsar-like powers. But all I can do is strongly suggest, and for most, it just goes flitting by.

My favorite backup tutorial site is very old, the Tao of Backup by Ross Williams (1997). Unfortunately it’s a little too long to get users to read it. It was written as an ad for now long-defunct verification software, but it’s still great and I’m glad it’s still available:

1 Like