[carterx]

I use NetBoot for fast deployment of our mages here at the college and when I tested Radmind a couple years ago I find it a bit hard to use. Just wondering if anyone has used the new 2008 Randmind version 1.1.1

Also, if you use Radmind, what would be the benefit of using it over Netbooting other than for cutting down time for imaging if it's only replacing certain files rather than whole images.

Thanks,

[Cold Warrior]

Radmind is an application, either via the command line or its GUI. Moving...

[larkost]

I think you are misunderstanding the point of Radmind. Radmind is not really a replacement for NetBoot/NetRestore, but rather a compliment to it.

At my job we have 400 computers that use NetBoot/NetRestore to get a base image on them, and then Radmind keeps them working the way that we want them (the correct apps, watches over and repairs things, allows for changes).

Here are two different situations:

When a new computer comes in (usually with 150 siblings) it gets NetRestor'ed with a base image that includes Radmind, and then told to Radmind (this is all after the network is told what the name should be, which means a lot here). At then end of that proccess the computer is read to go. I am looking into having Radmind creat .pkg's for me for this cycle so that my base image is close to being what I already want it to be.

But when a new patch comes out for the OS, or there is a new version of an application that we want, then it is Radmind that does the job. All of the computers check in with the Radmind server at night (staggered beginning at midnight), and see that I have made changes to the command files. They pull down the new command files, and start to work out the differences. In the morning all of the computers now have the new software on them.

See... not competing products, but two parts of the same setup.

Oh.. and it is 1.11.1.

[Peter]

got any more documentation on your radmind setup? Sounds good!

[larkost]

Both the NetBoot/NetRestore and the Radmind setup are pretty vanilla (well... we have some things that are specific to our setup, but those are details you would not need to setup your own install. The two parts are pretty independent of each other.

[CharlesS]

Here's what we did (keep in mind this was several years ago - there might be easier facilities for doing some of these things nowadays):

1. Set Energy Saver on the machines to have a scheduled startup at sometime in the wee hours of the morning like 2:00 AM (so if someone's turned the machine off, it still gets updated when it needs to)

2. Using crontab, set the machines up so they run a shell script of our creation, staggered, starting about 15 minutes after the time we set the scheduled startup for

3. The script did the following:

a. Try to gracefully quit all applications. If we can't (i.e. a user has left a document open overnight, and didn't save changes), we pop up a dialog box saying that we weren't able to update the system, because some application failed to quit, and we abort the update process.

b. Pop up a dialog box warning that the update is about to occur and that when it's done, it's gonna shut down and all unsaved work will be lost (just in case someone's pulling an all-nighter and is still in the building for whatever reason). Give a Cancel button in case a user wants to stop the update. Waiting 30 seconds is equivalent to clicking the "OK" button.

c. Start up a custom Cocoa app we made to pop up a dialog saying something along the lines of "This computer is being updated. Please don't try to use it" with a barber-pole progress bar in it. Note that we did not use iHook, because this app gets launched by a script that's running as root, and OS X has a security hole where non-root users can send the "do shell script" AppleScript command to Carbon and Cocoa apps running as root, allowing basically any process on the system to have root access. The custom app we made had code in it to throw away all Apple events it received (note: iHook may have solved this issue by now - I did report the issue to the developer a while ago, and sent him some code snippets, but since I have not tried the new versions, I can't guarantee that he implemented my recommendations).

d. Do the Radmind updating stuff

e. If the update went well, shutdown -h now. Otherwise, pop up a dialog box letting the user know that the machine might have problems and that he/she should e-mail us and let us know.

The script also had a bunch of stuff tailored to our specific needs, as well as some stuff to keep the user folders looking like we wanted them, which I won't bore you with.

What you need to have is two machines in your office - one to make the loadsets on and another to test and make sure they're working when a client pulls them. If a user says:

User: Hey, there's a new version of Firefox out, can we have it on our machines

You: Sure, it'll be on your machine when you come in tomorrow morning

Then, you just install the new Firefox on machine #1, upload the loadset to the server, then run the update script on machine #2 to make sure it pulls the update successfully. If it does, you can sit back and relax, and in the middle of the night all the rest of the machines will download the loadset and have the new Firefox by the morning.