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

Task and Job time "limits"

Created: 28 Sep 2010 | 7 comments

OK, so everyone probably knows how you can give a task / job a time that it will auto fail after (end task afer ____ minutes).

 

However, how does that work when you chain jobs and tasks together?

 

Say I have:

Job 1

--task a

--task b

--task c

--task d

--job 1a

-----task 2e

-----task 2f

-----task 2g

--task e

Task Z

 

Is the time determined by the root job / task? Is it compounded based on the tasks within the Job? or is it just based on the current task running?

 

IE if I have Job A runs 5 tasks, and is set to end after 100 minutes, but all 5 tasks are set to 10 minutes each... would the job end if any of the tasks take longer than 10 minutes, would it only end after 100 minutes, or would it end after 150 minutes.

 

I am pretty sure it isn't based on the actual tasks time as long as that task is within a job, as I have a job that has a time limit of 240 minutes, and a task that is in that job that is 120 minutes.  That task showed 140 minutes for completion, so it clearly wasn't based on the tasks time.

 

Thanks,

Comments 7 CommentsJump to latest comment

mclemson's picture

I thought it was the way you described, but your test suggests otherwise.

 

Perhaps if a Job has a timeout of 4 hours, and the job contains two tasks with an hour timeout each, it knows enough to give them the extra time available?

Mike Clemson, Senior Systems Engineer, ASC
Intuitive Technology Group -- Symantec Platinum Partner
intuitivetech.com

Thomas Baird's picture

The reality is that you have several timeouts.  One is the tasks themselves, which all have their own time outs.  Another is the job, which has it's own timeout.  What I've never tested is if there are 3 tasks that each take 30 minutes, and a job set to time out in 40 minutes, will task 3 even fire?  I suspect not.  I suspect that tasks 1 and 2 would fire, the job would fail, and task 3 wouldn't even fire.  But that's my current guess, not something I've tested.

And you're right about one thing - the time-out is ball-park, not precise.  Here, I suspect there is a scheduled task that evaluates tasks to see if they've expired or not, rather than a constantly monitored clock.  For instance, if ever 5 min something looked at task instances and killed any over their time, you might be at 63 minutes, or 61, or 64 for a 60 minute task.

The thing that impresses me about your find though is that a 120 minute task reported completion at 140 minutes.  It's almost as if the job timeout over-rode the task timeout.  hmmm...

Thanks for this submission.  Food for thought!

Thomas Baird
Private Consultant & open to full-time opportunities.
That means I CAN help beyond the forum (directly).

 

chris.vanderlinden's picture

It almost seems like the time limit of the Job will overrule that of any tasks within it.

 

However, I haven't had time to actually test that yet.

 

Thomas Baird's picture

Agreed - that's worth testing.  I wish I knew.  I'll add that to my to-do list. 

Thomas Baird
Private Consultant & open to full-time opportunities.
That means I CAN help beyond the forum (directly).

 

luisdg's picture

So...

Any update?  I can't believe we don't have an answer to this.  It sounds like something the devs would know for sure!

Thanks!

Thomas Baird's picture

Running through a SERIES of tests similar to this to discover how this all plays out.

My current "guess" is that Dev tested "expected" behavior, not unexpected behavior, if that makes sense.  Normally, you'd expect someone to build a timeout for a job that is longer than the cumulative timeout for the tasks in the job, so my guess is that's how testing and planning happened.  Thus, they may not actually  "know" what will happen in something like this.

 

Just a guess.  Either way, we're testing pretty thorougly now.  Learning a lot too!  Some new KB's will be forthcoming about things we're discovering.

Thomas Baird
Private Consultant & open to full-time opportunities.
That means I CAN help beyond the forum (directly).

 

mclemson's picture

Be sure to post back those KBs once they're posted.  I don't think there's a way, any longer, to subscribe to KB changes for Altiris products.

Mike Clemson, Senior Systems Engineer, ASC
Intuitive Technology Group -- Symantec Platinum Partner
intuitivetech.com