Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

DS 7.1 - Single Task Sever Job - Cannot Peform Full Deployment Automation By Itself

Created: 29 Jul 2010 | 2 comments
GEiST23's picture
  • So, I ran down the road of DS 7.1. Lots of potential, less architecture needed...the wave of the future and all those good things. And I believe it is the way to go for the future.

    However, we have one major problem right now and it is the biggest "miss the mark" in history of deployment software in my mind.

  • Create a Client Job
  • Add the following tasks:
  • Boot to Automation: PXE
  • Deploy Image
  • Boot to Production
  • Quick Delivery Software: App A
  • Quick Delivery Software: App B
  • Quick Delivery Software: App C
  • Quick Delivery Software: App D
  • Quick Delivery Software: App …etc.
  • At first approach you think: Yes, this will work and this be the answer I was looking for…great!

 

Then you find out two things about brand new computers found during initial deployment or imported CSV computers:

  • If you assign the client job to initial deployment = job fail
  • If you drag and drop the job on a computer that has reported into WinPE as minint…etc. = won’t allow you to schedule it ( a license error that when you KB it you find that it knows the correct plugins aren’t installed)
  • Why? Because there is a logic written that prevents jobs from running tasks that are not supported by the NS server’s knowledge of the plugins installed on the client. And of course in WINPE there is not much but the pectagen at that point so it is true. But, once it boots to production after the syspreped image is laid down with all the plugins this is a mute point. The very fact that boot to production task is listed in the sequence should loosen up the rules to allow software installation jobs to run in production.
  • And yes, there are work around and I have thought about them:
  • Copy all the packages in via WinPE once the images is laid down and then schedule scripted tasks to execute the software packages
    (could I invent more reasons to not use my software repository)
  • Use a SLOW managed software delivery chained software install
  • Don’t even think about targeting all computer and using detection rules – this doesn’t work for so many companies for so many reasons
  • So this is my appeal: One Altiris task server job get it done from soup to nuts…..end of story. Please let me know when we can get there or if I missed something in my DS 7.1 adventures that allows this to happen.

Comments 2 CommentsJump to latest comment

vortex0007's picture

I have to disagree with your idea that Task Server misses the mark for deployment software. From a software development perspective, Symantec is tasked with developing a framework and a set of tools that can be adopted to thousands of different business models and different needs. The Task Server framework accomplishes this quite well by allowing each task to be created independant of the rest of the job. Customers can then build jobs that chain together the various tasks that are unique to their organization.

Using your task server job as an illustration of my point, you use PXE, but many organization do not. If Symantec shipped 1 job that did it all, and that job forced customers to use PXE we would have a problem with those that don't use PXE.

This approach introduces a bit more configuration effort on the part of the administrator up front, but in the end you have complete control over your overall process. In other words - the tool can be adopted to fit your specific and often unique business process.

As I read through the list of problems you are having with Initial Deployment I know that some of these are being addressed in upcoming Hotfixes for Deployment Solution 7.1. I don't know when these are coming, but rest assured development is aware of the challenges you are having.

 

bering's picture

I have to agree with Geist23, I have the exact same challenges with DS 7.1 and the task server so much potential... But the somewhat simple task of deploying and image and some apps is way to hard!

Using managed delivery you get the question that looks good - but when is it done? And why is that when the agent restarts upgrading itself the delivery jobs does not continue until the next schedule kicks in....

Looking forward to those hotfixes!