Mac Server with DeployStudio. (really need some solid guidance)

[jonathanml86]

So long story short, I work for a company that purchases hundreds of Macbook Pro’s at a time. My job is to take usb thumb drives with the OS X install > wipe the current HD, and write a fresh OS X to the laptop, and then set it in the “ready to be sold” pile.


Needless to say, this method worked fine. But as I needed to reformat/fresh install OS X’s at a larger volume per week, I looked into better methods.


So I downloaded and installed the latest version of Apple’s ‘Server’ application. Also went out and bought myself a 24 port switch.


I have 2 Macbook Pro’s acting as the ‘blueprint’ and ‘admin’ machines. The ‘blueprint’ machine is a fresh 10.10.5 mac OS X. So I went into the System Preferences > Startup Disk > Target Disk Mode. Essentially, this machine is the machine I’m capturing my image from. It doesn’t have Apple’s ‘Server’ application installed on it, unlike the other machine, ‘admin’. I don’t think our end customers would want the ‘Server’ app. So this machine has restarted in Target Disk Mode.


>>>>Plug in a thunderbolt cable into the ‘blueprint’ machine, into ‘admin machine’>>>>


I ran the Disk Utility to ‘Repair Permissions’ on the mounted drive which is the ‘blueprint’ machine. I then ran System Image Utility, created a ‘Netrestore Image’. The source was the mounted drive which was the ‘Blueprint’ machine, I chose to customize the creation, added user account (so all would be default username and password). I saved the creation to a different location than the default (library/NetBoot/NetBootSP0). Because I decided to host this image on a SSD 4 bay OWC Thunderbolt 1 drive. So I formatted the drives, and then created my own Library/NetBoot/NetBootSP1 folder, and used this folder as the save to directory.


Process completed fine, had 23 machines ported into the switch, using the option at startup method to boot from hosted netrestore image(s). Install works fine on about 3-5 machines, but any more than say 5 machines, the install processes ALL FREEZE. Ideally, I want to have the 'Admin' machine plugged into Port #1, and Ports 2-24 will get a new machines as fast as I can clean wipe/os x install/


People had been recommending Deploy Studio, so I installed it, set up a DeployStudio Server, even “Created a DeployStudio NetBoot set”. Which I set up, and it worked. Then I realized this is a boot method to RUN the ‘Blueprint’s image it captured through the network. Versus what I need it to do, which is to INSTALL the ‘Blueprint’s Image’. So I can unplug it from the network, and there won't be any previous personal data on there, and it will be a fresh install of the os x, ala the 'Blueprint' machine.


I need a method that won’t freeze up the 23 machines. I want to plug in a fleet of 23, power them on holding Option, select the NetRestore Image to install on the 23’s internal harddrives, go through just the necessary prompts, which should be 1.) Choose English as default language, 2.) Confirm the hard drive listed is the install directory for the Netrestore Image.


I want to come back in the room, all machines are at the log in screen, ready to be used for the first time, I power down the machines, grab 23 more and do it again, and again and again. These machines will never be hooked up to my switch again.


Thanks to all who read this, and thanks in advance for any input and thoughts.


Cheers guys and girls!

Jonathan

 

[jonathanml86]

Buehler?

 

[chrfr]

I want to come back in the room, all machines are at the log in screen, ready to be used for the first time, I power down the machines, grab 23 more and do it again, and again and again. These machines will never be hooked up to my switch again.


Thanks to all who read this, and thanks in advance for any input and thoughts.


Cheers guys and girls!

Jonathan

You need to first use DS to create a master (it has a default workflow to do this), or you could use a tool called AutoDMG to create a base OS X installation, then use a DS workflow to install your other packages after the OS is copied to the disk.
It's worthwhile to take the time to read the DeployStudio documentation.

 

[jonathanml86]

So I got a little further...

I used DeployStudio Admin > workflows

1.) I made a master, source was my clean os x install machine 'Blueprint' booted into Target Disk Mode
-this went fine, it put itself in the HFS directory within the DeployStudio folder

2.) I then made a custom work flow. Used the 'Restore' function, set the HFS folder, and selected my master dmg I created in the above step.
-this went fine, I then ran 'Runtime'

This is where I'm scratching my head. Does DeployStudio need to be installed on every single machine I plan on re-imaging? That would create a lot of unwanted time for my process. I am trying to find a great/consistent alternative to using the NetRestore Image process. Which was wiping out a machine's current HD, and installing my image of a fresh os x install. Again, this older process was freezing up when only using the Netrestore feature within Server. Again powering off a machine > OPTION + Power on, selecting the Netrestore Image from the startup disks. It will work fine on 1-4 machines, but anything more would freeze the image install process. Thought Deploy Studio would solve this, but right now I'm under the impression while yes, DeployStudio could work, but I would need the Deploy Studio software on EVERY single machine I intend to reimage???

 

[chrfr]

but right now I'm under the impression while yes, DeployStudio could work, but I would need the Deploy Studio software on EVERY single machine I intend to reimage???

Read the documentation. You need to either create a DS Netboot or create a DS runtime on an external disk. http://www.deploystudio.com/Downloads/Extras/DeployStudio_Guide_2.0.pdf Has the information you need, beginning on page 29.
You must use the DeployStudio netboot image or external drive to do what you need.

 

[jonathanml86]

So I plugged in a machine to my switch, did the OPTION + Power on

My deploystudio option is present, used that as the start up disk > saw the Runtime menu!!! Selected my customized workflow!

Testing now!

 

[jonathanml86]

So the install/reimage process but it had an error and didn't complete. It said to check the log (on ther server machine) so I am in the log in the DeployStudio admin program. My log photo is attached.

 

[jonathanml86]

Read the documentation. You need to either create a DS Netboot or create a DS runtime on an external disk. http://www.deploystudio.com/Downloads/Extras/DeployStudio_Guide_2.0.pdf Has the information you need, beginning on page 29.
You must use the DeployStudio netboot image or external drive to do what you need.

I did create the DS Netboot is the assistant right after I created the DeployServer

 

[jonathanml86]

Tried it on another machine, success! (seemingly so)

 

[jonathanml86]

Ok so everything I did last, worked. Granted at the time I only had 1 machine to test the deploy on. But nevertheless, it worked. Reimaged/Restarted, on to the next one....However, I did this trial run on my personal machines, but I took my notes to the office, to do the EXACT same thing. Went step by step, NO room for any mismatched variables, except the actual computers I was using as the 'admin' and the 'blueprint'. Same model of Macbook Pro.

When I boot up even just one computer into the DeployStudio Netboot, it gets to the Deploy Admin screen where I select my custom workflow (which is just a restore with multicasting, skip recovery, automate.

I get a long list of log error (which I will attach below) and it ends up failing, and the machine ultimately restarts to a blinking question mark white screen. So I at least know it's 'wiping' the hard drive in the process, but not reimaging the machine(s).

Can someone give me a solid idea as to why (if I used the same guide in my setup) it would have log errors and fail with the only variable being, different computers for the 'admin' & 'blueprint'. I beat my head against the wall for nearly 10 hours today, and 30 over the weekend. I am needing some good veteran/pro help!

Thanks in advanced guys

PS, there is much missing in the screenshots between image 3 and image 4. It was all the same-ish log, so I didn't include it because you can see some of it in both 3 and 4.

 

[Silencio]

You might want to read the DeployStudio forums:

http://www.deploystudio.com/Forums/index.php

It's a great product, though some of the setup steps could be more streamlined and made a little more clear. Nothing that a little RTFM can't fix, at least in my case. It's freeware, so can't complain too much.

I've barely even scratched the surface of what I want to do with the workflows, but I'm a bit short of free time to work on it more.

If you're re-imaging a lot of machines at once, set up multicasting on the server side. My deployment server has a 4x1GB ethernet link aggregate connection to our core switch. It's pretty rare that I find myself re-imaging more than one machine at a time unless I'm rolling out a whole new image to everybody.

 

[DJLC]

Looks to me like you might need to adjust your multicast settings (disk speed and network speed) in the DS workflow. I know when I do a multicast restore I see the same errors, but somehow it usually works. It's just highly dependent on those settings. You might also check your switch's options assuming it's a managed switch.

Failing that, try turning off multicast and see if it is successful with unicast. That will help you narrow it down.

Sounds like you've otherwise done things right...
1) Install DeployStudio on Server.
2) Create DS NetBoot master.
3) Boot "blueprint" computer to DS NetBoot master, use the "Capture" workflow.
4) Boot the "target" computer to DS NetBoot master, use the "Restore" or a custom workflow.

I will say I typically am only doing 7 machines at a time (we have MacBook Air, and I only have 7 TB->Eth adapters), but with those 7 I honestly don't notice much difference between multicast and unicast. Obviously unicast isn't as efficient, but it sounds like you're doing this on an isolated network segment so it might be worth a shot.

 

[jonathanml86]

Looks to me like you might need to adjust your multicast settings (disk speed and network speed) in the DS workflow. I know when I do a multicast restore I see the same errors, but somehow it usually works. It's just highly dependent on those settings. You might also check your switch's options assuming it's a managed switch.

Failing that, try turning off multicast and see if it is successful with unicast. That will help you narrow it down.

Sounds like you've otherwise done things right...
1) Install DeployStudio on Server.
2) Create DS NetBoot master.
3) Boot "blueprint" computer to DS NetBoot master, use the "Capture" workflow.
4) Boot the "target" computer to DS NetBoot master, use the "Restore" or a custom workflow.

I will say I typically am only doing 7 machines at a time (we have MacBook Air, and I only have 7 TB->Eth adapters), but with those 7 I honestly don't notice much difference between multicast and unicast. Obviously unicast isn't as efficient, but it sounds like you're doing this on an isolated network segment so it might be worth a shot.

I did use the multicast, however I did not adjust the parameter stuff.

This network doesn't see any other activity beyond deployment/reimaging. I use a D link 24 port switch. I am kinda new to this deployment stuff. Any further insight on this?

 

[DJLC]

I did use the multicast, however I did not adjust the parameter stuff.

This network doesn't see any other activity beyond deployment/reimaging. I use a D link 24 port switch. I am kinda new to this deployment stuff. Any further insight on this?

I'd start by dragging the sliders down to maybe half their current setting and see what happens. If it's successful, you can try moving them upward in small increments until you start having problems again. As you experiment you'll kind of get a sense of your sweet spot.

Otherwise, like I said, it looks like you're doing everything right.

 

[jonathanml86]

I'd start by dragging the sliders down to maybe half their current setting and see what happens. If it's successful, you can try moving them upward in small increments until you start having problems again. As you experiment you'll kind of get a sense of your sweet spot.

Otherwise, like I said, it looks like you're doing everything right.

I didn't move them at all from their preset values that became enabled, as I enabled the multicast. So even the low settings they are at by default, are too high?

 

[jonathanml86]

I think I may have done something wrong in the beginning. So I had 2 machines. 1 was my 'admin', had the Server app installed, and DeployStudio. I set up the server, then the deploystudio server on this 1st 'admin' machine. Then I created the DS Netboot set from this same machine HOWEVER, @ the 'Source Base System' selection, I chose my 'blueprint' machine that I had in Target Disk Mode connected via Thunderbolt.

Is this step messing me up???

Should I be:

on the 'admin' machine install Server, Install DeployStudio, and set up its server

Then >

running my 'blueprint' machine, install Deploy Studio, and then create the Netboot set from here, but place it to the server for use?

 

[DJLC]

I think I may have done something wrong in the beginning. So I had 2 machines. 1 was my 'admin', had the Server app installed, and DeployStudio. I set up the server, then the deploystudio server on this 1st 'admin' machine. Then I created the DS Netboot set from this same machine HOWEVER, @ the 'Source Base System' selection, I chose my 'blueprint' machine that I had in Target Disk Mode connected via Thunderbolt.

Is this step messing me up???

Should I be:

on the 'admin' machine install Server, Install DeployStudio, and set up its server

Then >

running my 'blueprint' machine, install Deploy Studio, and then create the Netboot set from here, but place it to the server for use?

To your first post: I forget what the default settings actually are. In my environment I've had success with a stream data rate of 30MBs and client disk speed of 125MBs. That's coming from an Xserve with dual bonded GigE to MacBook Airs with Thunderbolt to Ethernet adapters.

As far as creating the DS Netboot set, I've always done it directly on the server. You shouldn't need to do Target Disk Mode at all. That said, I'm not sure that it matters... I think it just needs a valid and complete OS X Install to steal files from. That said, make sure you're creating the DS Netboot with the newest version of OS X you use to avoid any issues booting different Mac models.