NetRestore Image

[jonathanml86]

Hello all,



So I've set up a server environment to provide a Mac Reseller the ability to use an image capture of a clean/fresh install of Yosemite 10.10.4.



I chose to use the NetRestore Image process within the System Image Utility service > saved to an SSD in my OWC Thunderbolt - 4 bay drive. I did this process 4 x's, because I am using the network drive independently (as 4 individual drives, rather than 1 aka RAID). I decided to the 4 separate, because our switch size is 24 total ports, so essentially deploying an image to 23 machines at a time. So I assumed 23 /4 ( each SSD drive) = better timing; versus 23/1 (previous set up was 1 usb 3.0 external ssd hard drive).



So on the previous setting of a usb 3.0 - SSD external harddrive, I did not make any partitions, however I did host another identical restore image within the machine's internal hard drive, to go along with the external ssd hard drive.

----We were deploying/restoring machines within 12-19 minutes, and crushed the amount we did.



Now that I'm on the new setup of a thunderbolt OWC brand, 4 bay, 2.5 SSD network drive.

- booting the machines into the network image files (again, i have 4, 1 for each of the SSD bays), the process freezes eventually, and only a few out of the 'massive' batch of machines ported in at that time, make it all the way through the process. I am really frustrated because we increased our technology but the results are worse...



Does anyone have any suggestions, or input?



Thanks!

 

[Yebubbleman]

Maybe I'm missing something here, but why didn't you just make a NetInstall image of the 10.10.4 installer?

Also, I might be wrong, but I don't think you need to replicate the image onto four different volumes to deploy to 23 systems. Unless you're on a really old or really horribly engineered network, OS X server shouldn't choke on imaging 23 client machines at once from the same single image. Multicasting is your friend...

 

[Yebubbleman]

In fact, you may very well be making the process worse by making it more complex than it has to be.

 

[MacsRgr8]

Maybe I'm missing something here, but why didn't you just make a NetInstall image of the 10.10.4 installer?

Also, I might be wrong, but I don't think you need to replicate the image onto four different volumes to deploy to 23 systems. Unless you're on a really old or really horribly engineered network, OS X server shouldn't choke on imaging 23 client machines at once from the same single image. Multicasting is your friend...

Mulltcast is the key.
Unless you can get a 10Gb ethernet card into an Xserve or cMP as the NetBoot server, and have a switch which has a 10Gb port, you must use multicast to be ably to restore images on more than, say, 10 Macs simultaneously and keep the server "happy".
The Gb interface of the server is the bottleneck when deploying multiple Macs using unicast (standard).
BTW... RAM can be an issue too. FileSharing the images can eat up the Server's RAM. Keep watching Activity Monitor.app.

Maybe http://www.deploystudio.com can help you too?
It's an excellent free tool to help you create and deploy images for OS X.

 

[Les Kern]

Mulltcast is the key.
Unless you can get a 10Gb ethernet card into an Xserve or cMP as the NetBoot server, and have a switch which has a 10Gb port, you must use multicast to be ably to restore images on more than, say, 10 Macs simultaneously and keep the server "happy".
The Gb interface of the server is the bottleneck when deploying multiple Macs using unicast (standard).
BTW... RAM can be an issue too. FileSharing the images can eat up the Server's RAM. Keep watching Activity Monitor.app.

Maybe http://www.deploystudio.com can help you too?
It's an excellent free tool to help you create and deploy images for OS X.

I'd personally suggest staying away from multicast. This looping everywhere does horrible things to a network. We wound up doing one image on one server using the old NetRestore software and did 30 machines at a time with no worries. Sure, it took awhile, but there was no network hit. Most times we'd set it to fire at the end of the day and let it image. At times we did 90 at a time using the same image. Agree with Yebubbleman... keep it simple.

 

[MacsRgr8]

I'd personally suggest staying away from multicast. This looping everywhere does horrible things to a network. We wound up doing one image on one server using the old NetRestore software and did 30 machines at a time with no worries. Sure, it took awhile, but there was no network hit. Most times we'd set it to fire at the end of the day and let it image. At times we did 90 at a time using the same image. Agree with Yebubbleman... keep it simple.

Of course, don't use multicast on a network that is actively in use by others! Multicast on a switch is like unicast on a hub: brings down all other networking traffic.

Image the Macs when others are out. Otherwise use unicast and wait.
IMHO, that is simple.

 

[Yebubbleman]

Maybe http://www.deploystudio.com can help you too?
It's an excellent free tool to help you create and deploy images for OS X.

While I like DeployStudio, I don't see it as being all that much better than simply using the standard OS X Server NetInstall service with System-Image-Utility-made images...UNLESS you don't want to actually store your images on the server from which you are deploying. Furthermore, given this, it also helps make your deployment server super easy to just rebuild from scratch in the event of a failure (as rebuilding it takes a comparable amount of time to simply backing it up and restoring from said backup).

 

[jonathanml86]

Thanks everyone for the input. So let me just double check, the fact that I was using an identical image, labeled 1, 2, 3, & 4, on a 4 bay, OWC SSD setup > hooked up to my server machine (macbook pro) via thunderbolt, was causing a bottleneck??

And to keep things simple, in theory, I only need 1 restore image hosted on the server computer and the deploy should run smoothly? This makes sense give the fact that my original setup was on a pro with a external ssd via usb 3.0 connection. And the deploying of the image went through many many times with no hiccup

 

[jonathanml86]

Seeking some more help guys...

Again my setup is

So now I am testing the waters in Deploystudio

1.) I've set up my server
2.) I went through Create a DeployStudio Netboot set
--which worked

I didn't know the method I used was to create an image for computers to use

Again, the environment of my setup is:

1, we get a mass amount of macbooks
2, we hook them up to our switch into our server environment, to get a wipe/fresh install of a netrestore image
--the first setup just using the Server app, was causing the macbooks to freeze midway through the reimaging process

So I was told to use DeployStudio
--while the steps I found worked, they worked for the wrong thing. I don't need this macbooks to just load the image I captured using the 'Create a DeployStudio Netboot set', I literally need that captured image of that machine, to install clean on these interchanging macbook hard drives. Afterwards, I unplug them, reboot them, make sure they are freshly formatted os x's for resale, then I plug in another machine that needs to get wiped, install a fresh os x using my restore image

Can someone please help me ?

Thanks in advanced

 

[DJLC]

Here's how I do this...

DeployStudio on OS X Yosemite Server. I installed fresh OS X on one MacBook-for-sale, booted into the DS NetBoot image, captured the fresh install of OS X. Then on other machines to be sold, boot into DS NetBook image and restore the fresh OS X image. Works perfectly. And depending on the version of OS X you choose you can use the same image on multiple models.

I prefer DeployStudio's workflow model to a simple NetInstall, but it's all down to personal preference.

I would keep your images on a RAID of some sort. One copy of the image. I typically only do ~7 MacBooks at a time, but we're all MacBook Air and I have a limited number of Ethernet adapters. Theoretically this should scale up with no issues. Be sure not to set DeployStudio's multicast sliders too high; this can cause failures. You may need to do some experimenting to find the "sweet spot" for your particular network environment.

Also agree with Yebubbleman; keep it simple.

 

[adam9c1]

In the current DS can you run a task to delete HD securily?

 

[DJLC]

In the current DS can you run a task to delete HD securily?

There is a workflow to enable / disable firmware security. You can additionally do a secure erase by using the "Partition" task (and set to one partition, or however many you might want). Or failing both of those you could create a shell script to do it and insert that as a task before "Restore."