Looks like no one’s replied in a while. To start the conversation again, simply ask a new question.

Netrestore of 10.11.4 and 10.11.5 FAIL

I've been using with success for quite some years netrestore to deploy our macs.


Ever since I upgraded the master image and the server to 10.11.4 the process fails. Experience teaches that usually this gets noticed and fixed. Alas 10.11.5 still has the identical same problem.

I can still deploy the 10.11.3 based image (but well it's a PITA to have to manually do all the upgrades every time).


Setup:

Mac OS X 10.11.5 + Server.app on mac mini server (tried multiple servers, even one on a clean install.

  • Netrestore image create with system image utility on and off of a 10.11.3 machine hooked up via thunderbolt in target disk mode: continues to work as it used to do
  • Netrestore image created exactly the same way on and off of a 10.11.4 or 10.11.5 machine: can't get it to work.


Clients tried: multiple MBP's ranging from a brand new machine to a 5 year old machine. All yield the same: progress bar under the apple logo goes to 70/80% and that's it.


Since the 10.11.5 server can still serve the 10.11.3 created image, it doesn't look like it's the server side, it looks like system image utility needs a fix.


If you are REALLY patient (multiple hours), it seems the machines just crash: hard power off is the only thing they respond to anymore - black display.


So the question:

Anybody got netstore of 10.11.5 based images working ? What did you do to work around the issue ?


It's going to be really bad as soon as Apple rolls out new hardware that might have a custom build or something like that before they fix this.

OS X El Capitan (10.11.5), System Image Utility - netrestore

Posted on May 26, 2016 8:32 AM

Reply
21 replies

May 26, 2016 8:58 AM in response to sss666

Well I have to add one thing: while typing up the above I did a few more tests, and tried on an older MPB I had sitting on my desk for other reasons to get it to boot off of the server a handful of times off of one of the servers. It failed as expected a handful of times, but I had left it hanging while editing up the question in here.


And somehow, miraculously it seems to have popped through somehow. So the failure rate is not 100% (but given the amount of tries it's so high as to be unusable in practice. But given the lack of 100% failure rate ... could be harder to debug than needed esp. since images fail to want to boot in verbose mode (or I don;t know how to trigger verbose boot during a netstore.

May 26, 2016 4:49 PM in response to S.SubZero

Might very well be the same issue between netrestore and netinstall. That other thread seemed to focus on connections: I've the problem with both thunderbolt/ethernet apple branded adapters as well as older MPBs that still have ethernet on-board. I've no usb to ethernet adapters.

FWIW: I tried in in 2 physical locations, so 2 different networks etc. all seems irrelevant.

May 31, 2016 3:10 PM in response to sss666

I have the same issue, I started with 10.11.4 then called apple support. They stated the issue is a known issue and was fixed in 10.11.5, I updated, tested and got the same results. I called Apple back and they forwarded the info to the Apple engineers, it sounded like they thought it was fixed but it's obviously not. They did not give me any info as to what the issue was but did say they had all the information they needed, meaning they know what it is and just need to figure out how to fix it. No ETA on a resolution.

Jun 1, 2016 6:20 AM in response to sss666

If you are fast enough you can also start a netboot in verbose mode to see why the machine hangs during the boot process.

I tested it here and it seems to me, that now even on real Macs the "Waiting for DSMOS" problem has occured at least with NetInstall. DSMOS = Don't steal Mac OS...

During my tests with a MacBook Pro 4,1 and a MacPro 5,1 and NetInstall of 10.11.4 and 10.11.5 I found out, that it could help, that, if the machine hangs and it is connected via a LAN port, to disconnect the LAN cable for approx. 5 seconds and then reconnect it. This seems to re-initialize the network adapter and its connection and in my case I could successfully boot with 10.11.5/10.11.4 NetInstall images - although it took very, very long in relation to Mavericks.

Of course, this is not a real solution for the problem. I also hope, that Apple fixes this soon.

Jun 3, 2016 7:21 AM in response to MacPro_de

****: Interesting!


First off: I'm only using true Apple hardware, never have I touched anything alike a hackintosh.


Yes, one can verbose boot a mac during a netrestore, but the firmware password needs to be removed (not just entered) to do that apparently.

[a bit unhandy and waste of time, esp. if the mac has no recovery partition on it]


Verbose booting side-by side a 10.11.3 and a 10.11.5 image on relatively old macs (Last 15" pre-retina MBP and one of even a generation older were sitting here waiting to be sold)


Both end up initialising their bluetooth interface and enter it into promiscuous mode and then the difference starts:


The 10.11.3 one prints:

AppleLPC::notifyPlatformASPM - registering with plugin with ASPM Support true

followed by a long wait and then indeed something like

DSMOS has arrived

where the screen right after that goes to the graphical interface.


The 10.11.5 one prints:

NVDAGK100HAL Loaded and registered

and then seems to hang forever.


I let it hang for long enough, and then tried to disconnect it a few times from the ethernet, but none made it go on at this point. Nor were therre any kernel messages of the interface going down any longer.


Tried a second time with the 10.11.5 one: disconnected and reconnected a few times before it hangs and indeed it booted properly in the end.


<Personal Information Edited by Host>

Jun 1, 2016 6:36 AM in response to sss666

[edit time is too short]
Tried a second time with the 10.11.5 one: disconnected and reconnected a few times before it hangs and indeed it booted properly in the end.

It did print the lines

AppleLPC::notifyPlatformASPM - registering with plugin with ASPM Support true

DSMOS has arrived

and a few more I didn't manage to capture yet after the

NVDAGK100HAL Loaded and registered

So seems it's the part that cause the AppleLPC::notifyPlatformASPM - registering with plugin with ASPM Support true to be printed that's causing the crash it seems.

And indeed a rightly timed dis/reconnect of the ethernet seems to do the trick.

Jun 3, 2016 7:20 AM in response to sss666

Thanks to the helpful input from ****, I've been able to create a workaround:


  1. Optional, but recommended: remove any firmware password the system might have.

    Boot it from either a recovery partition (if there's one), an external medium, or an OLD (10.11.3 or earlier netboot/netrestore image) to get to the tools to do that

  2. Boot the system from the 10.11.5 netrestore image in verbose move from a fixed ethernet (using a thunderbolt to ethernet dongle if needed)
    Verbose mode is actually optional, but it helps a lot in your timing.
    • Hold OPTION
    • Select the image using the arrow keys
    • Hover fingers over command and V
    • press ENTER immediately followed by pressing command-V
  3. At the right time disconnect the ethernet and reconnect it
    The reason and exact timing still eludes me - but it works for now ...
    • Wait long enough for the system to have mounted its new root partition - if you do it too soon, some systems will panic (retina MBPs like to do this it seems)
    • The kernel will print disconnect and connect events - if it does not anymore: you waited too long, system is already frozen- hard power off and try again
      This is also why it helps to have it in verbose mode otherwise you don't know when it froze
    • You can do it a few times - wait a bit in between.
    • Unplug the ethernet cable if you're using a thunderbolt dongle: unplugging the dingle itself seems to trigger kernel panics quite easily.
  4. Remember to add the firmware password again if you removed it.

    And avoid making the same typo twice ... you do not want to "forget" that password.


Once the system switches to graphics mode: you're done, just a short wait and it'll be ready to go on as it should have without all this hassle.


I did try other combinations:

- succes, but harder to get it right on first looks: let it netrestore via WiFi with an wired ethernet connection plugged in, and unplug the ethernet at a given point (unfortunately it's still was time critical and has the added hassle of having to connect a system to the WiFi as well as it being slower that way.

- no success: using wifi only


To me this workaround allows me to image 10.11.5 instead of 10.11.3 and then upgrade all the stuff manually at the cost of not being able to do this "hands-off" anymore. The point where this is easier or not will be different for every individual and will greatly depend on volumes and/or level of non-apple sourced updates one needs etc.


I for one would appreciate more insight on timing (when's too early, when's too late for it to have effect) and what exactly this reset of the ethernet interface does to allow the machine to pass the copy protection of OS X (all on genuine Apple hardware).


In the unlikely case Apple read this: It's genuine Apple hardware, we didn't pay you to be bothered by DSMOS crap, let alone you dragging your feet to fix this.


<Personal Information Edited by Host>

Jun 3, 2016 10:36 AM in response to sss666

Just for clarification: I don't know for sure, if DSMOS really is the reason for the problems - I only think, that it is one possible reason.

It could also be a problem with a driver kext and/or with SIP.

As Apple stated, they are aware of the problem with NetBoot, so let's keep our fingers crossed, that they will solve this problem soon.

In any case I'm glad, that my test results with 10.11.5 NetInstall also allowed you to find a workaround for your NetRestore problems.

Jul 7, 2016 11:31 AM in response to sss666

I'm having same issue, the base 10.11 installer seems to be the last working installer which is nowhere to be found. Also the latest Server App / Image Utility creates a new kind of net install. I created an automated installer which worked on iMacs but cause Macbook Pro's to freeze and get stuck on "considerRebuildOfPrelinkedKernel rebuild rebuild has expired"


Boot the machine in verbose mode to get the above. Hoping that Apple can come up with a solution asap as we need to upgrade a great amount of machines..


Also using the latest disk maker to create an installer works when upgrading from USB installer. It looks like it's not going anywhere with the "Apple logo loading bar" but eventually reboots and completes the install in about 30mins.


Had to revert to USB installers to get machines upgrades 😟

Jul 27, 2016 6:08 AM in response to sss666

HI,


this is a very weird problem, I created a 10.11.6 NetRestore Image, made in a 10.11.6 Mac to be deployed on a 10.11.6 Mac, and still the problem persist. (to see if this works finally after 1000 test with the 10.11.4 and 10.11.5 updates)


It doesn't matter how much you try it always end on the Apple logo going until the middle, take some more minutes and then crash hopeless, you go to the logs on the Server and there is no offer passing, Discovery and Request and some AKC and that's it... logs do not give more information about the error, It will be good if you add something like "Unable to contact the Computer", "Client is running a old version please update" or "Please add the server in the bootable list on the computer".

I can blame Apple and his new System Integrity Protection. even If you add the server with the csrutil netboot command this is not functional and is creating problems for everybody, good feature by the way... is like the Administrator password request is not enough security after all.


I will try to disable csrutil and see if this works.

Netrestore of 10.11.4 and 10.11.5 FAIL

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple ID.