Originally published at: https://tidbits.com/2019/11/06/dont-interrupt-security-update-2019-001-mojaves-installation/
Reports of boot problems after installing Security Update 2019-001 (Mojave) are rolling in. The common factors seem to be installing on a Mac with a T1 or T2 security chip and the user interrupting the installation because it seems to take too long.
I think it would be helpful when the installation process was accompanied with info for the user on what’s going on. Apple could for once tone down the minimalism a bit and prevent this kind of thing by letting the user know that the next bit in the update may take a long time and that it is to be expected. Why not?
More progress feedback is always good, except for the fact that it’s notoriously incorrect and variable between devices. Even with the install that I tested for this article, it first estimated 34 minutes, but quickly changed that and ended up finishing in 10. And if it really does have to do some of this stuff while the screen can’t be on, due to the low-level nature of what’s happening, it would be hard to provide progress feedback.
Nevertheless, I agree entirely that Apple should warn first that interrupting the installation could leave the Mac in a problematic state whenever low-level bits are being twiddled.
Feedback about device status has become worse with time, I think. The power light on the new Mac mini 2018 doesn’t even ‘breathe’ anymore when the Mac is sleeping, it just keeps shining like the thing is on. But I digress.
The Apple TV power light blinks when an update is taking place, the same could be done on a Mac. That, combined with a clear description and warning before the update is started, should prevent a lot of issues like these.
1 Like
I agree with this wholeheartedly.
And the fact that Apple, the computer company with $250B+ in the bank, in 2019 still can’t get a decent estimate for remaining time to install is frankly pathetic. We used to make fun of Windows’ silly estimates. Well, nowadays the Mac has reached about the same level. User feedback is one of the most important aspects of HMI. Embrace it. Tell the user what you’re doing, how long it’s going to take, etc. If you have lights and sounds, use those. If you are going to lose display output or lights because of low-level operations, tell the user that is about to happen, tell them how long that will take, and then tell them once you’re done (or if you’ve failed). This is not the place to be terse. This is IMHO an aspect where Ive minimalism has no place whatsoever. Ideally, everything will run smooth and fast and in that case nobody will be forced to see/hear much about it at all. There’s your minimalism. But not by removing feedback.
2 Likes
The problem being reported is a black screen for an unexpectedly lengthy time. It would appear that this is due to the inability to display anything on the screen of modern Macs during the installation of firmware of some sort. Now that some Macs come with T2 Security chips, I would have to guess the updates for those are the cause, but my relatively new iMac without a T2 has occasionally been known to display that lengthy black screen, so that can’t be the entire story.
Since I’m used to that now, I always let it sit for an hour or more if an update goes on for a long time.
This happened to me. The update was beach balling and seemed completely stuck. Foolishly I interrupted it, and it seemed to finish. But the APFS volume was toast. It was not possible to log in, despite the users being present. Recovery didn’t work, as you couldn’t decrypt the volumes, and so they didn’t appear as a target for recovery. Basically it required complete reinstall.
2 Likes
Ouch, sorry to hear it. The problem is that we just had an example where you had to force-quit an install, with the initial release of Catalina, so we as a community are not accustomed to interruptions having such significant repercussions.
Thanks at least for warning us Adam. If Apple isn’t going to take proper care of its users, at least we can count TidBITS to do so.
1 Like
Right on Brother!
Beyond that we’ve totally lost the startup chime that lets you know that the machine is at least waking up. I understand that chime (Boiiing!) would disturb situations in meetings and classrooms where people were firing up notebooks and such. The least Apple could have done was put an option to switch it on or off instead of just ditching it altogether. That was part and parcel of the Mac core experience. It’s not like eliminating a piece of hardware (the ⅛" audio plug on the portables) as the software to make that happen should be just about insignificant in cost on a per machine basis.
Apple had that option built in.
If my Mac had been muted before I started it, the chime would be silenced.
Even in the event that I’d forgot to mute it, or if I shut it down before knowing I’d be starting it in a meeting (not that I ever really shut my Mac down these days) where I’d want it muted, plugging in headphones would also silence the chime. The exception here I’m aware of is on an older Mac mini where the built in chime would play through the internal speaker regardless of headphones being plugged in or not. :(
What I do is disable any scheduled shutdowns, set Sleep to “Never”, then start the Security Update before I go to bed. I wake up in the morning to a completed update and reset my Shutdown & Sleep preferences to their normal settings. However my MacBook Pro with Mac OS 10.14.6 installed hasn’t mentioned any Security Updates recently. When was this SU released?
@ace, you repeated the recommendation last week to not upgrade to Mojave due to concerns with Mail etc. Is that still the case, or we OK to upgrade now (as long as we don’t interrupt the process)? Thanks!
Yes. This should have been in the article. It’s precisely what I came here to say. I mean, common sense says don’t interrupt. But then you sit and watch for half an hour while nothing happens, and finally start googling on your phone and find that you need to interrupt. I’m a lot more inclined to interrupt after that experience.
I believe the warning was not to upgrade to Catalina until the loss of mail issues were deemed addressed. Right?
We still recommend holding off on the upgrade to Catalina due in part to concerns with Mail. That doesn’t affect Mojave.
Yeah, I debated mentioning the Catalina interrupt issue there, but decided that it was just muddying the issue to have in the main text. Especially, as we can see in the comments, some people are having trouble keeping the versions straight as it is. That’s another downside of all the updates—it’s enough to make anyone’s head swim.
As you say, we know better than to interrupt installations willy-nilly, but we also all do it on occasion when something is going wrong. That’s why I was talking about the BridgeOS issues and the T2 chip—those really change the calculus when updating, since anything that changes them is doing more than just writing files to the boot drive.
1 Like
My bad - sorry!
I have yet to see any software update or download that is remotely accurate in its time-to-completion prediction, on either Mac or Windows (including, of course, OS updates). Part of the reason, naturally, is that the actual download speed is constantly in flux, in the case of my Internet connection, majorly in flux - might be 150kB/s for 30 seconds, then 1Mb/s for 30 seconds, then 500kB/s, etc. (according to both iStat and Folx), on all devices. I’ve long thought that all developers should just ditch the time-to-completion estimate altogether, just use the status bar graphic and maybe a real-time speed display so you can see it’s actually downloading. Not possible while doing a firmware update, but generally.