10.13.3 Suplemental Update failure loop

[jason_bracy]

So about 25% of machines that are suffering from an error when applying the "macOS High Sierra 10.13.3 Supplemental Update" released today.
"macOS could not be installed on your computer"
"The path /System/Installation/Packages/OSinstall.mpkg appears to be missing or damaged. Quit the installer to restart your computer and try again"

Restarting just brings it back to the same screen. the only fix I've found is to restore to a snapshot.

Anyone have any ideas as to why this is happening?

[mbracco]

We have this since 10.13.0 release. Every update that comes 10.13.1, 10.13.2, 10.13.3 all supplemental updates, we have this issue all the time on yes lets say also 25% of our imacs. 10.13.3 was a little better but still some failing.
We let the users boot to recovery open terminal and delete (rm -rf /Volumes/Macintosh HD/macOS Install Data) and with startup selection boot to Macintosh HD. A second try installing the updates worked mostly. Some infos we found is that connected USB-Sticks, thunderbolt adapters, USB-C devices makes the updates fail, apparently to protect those devices during the firmware updates apple does. We got the good advice to take off all connected devices before updates but as we use on all computer a DVI-VGA Adapter, this is not possible.
So hopefully future updates will improve and get as in 10.12.x an before a near 100% no touch update success.

[j_tanudjaja]

Confirming that this happens to us for more than 50% of the users. We told them to:
1. Click on the “mountain” image in the middle of the screen so that the apple icon is visible on the top left corner
2. Click Apple Menu (Top left corner of the screen)
3. Click StartUp Disk
4. Click Restart

[jason_bracy]

Thanks for the insight. I had not seen this issue on any of the previous updates, and couldn't find any info anywhere about what could be the cause. I appreciate the help. Just disconnecting any connected devices seems to allow the update to complete.

[cpoag]

Does anyone know if this issue with 10.13.1, 10.13.2, 10.13.3, etc. is a general issue or an issue specific to computers managed by Jamf? We're also being hit with a very high percentage of computers having this issue. We were about to send out a notification across all of campus telling users not to load any of these updates, but we're now wondering if there is a better approach.

[dgreening]

Tobe honest I have 100% given up on installing dot updates (combo etc.) on 10.13.x via Jamf delivered pkg. Look to automate the "softwareupdate" binary to do your updates and you will be far, far happier.

[jason_bracy]

We actually do use the “softwareupdate” binary triggered by a policy. And still receive the failed update on an alarming number of machines.

[mschroder]

We have seen this the first time with the 10.13.4 upgrade. The easiest to get out of the loop is to restart with option key pressed, then select the normal system disk. The system is untouched and you can boot normally. You can then just attempt another update, usually it works out. There is no need to reinstall the Mac, boot from Recovery, go back to latest snapshot or whatsoever. Just boot with option key and select your system disk.

In the beginning I also removed all traces of the attempted first update, thinking the update got corrupted on transfer, but comparing it to a good update I did not see any diffs, so now I just skip that step.

It very much looks as if the image the machine is set to boot from does not exist. I wonder whether user shut down or rebooted the Mac during the updates. The confusing thing with these updates is that the screen goes completely dark multiple times during the update, and on newer Macs you have no indication whatsoever whether the Mac is doing something or whether it is off.