Automatically handle activities that fail

[sumitgulati]

Hi All,

I have a Sequencer Job that calls a Server Job. I want the Sequencer Job to abort/throw warnings if the called Server Job aborts/throws warnings. I know one way to implement this is to call a routine Utility????ToLog also from the Sequencer Job. But since I am on version 7.1 I checked the option "Automatically handle activities that fail" in Sequence tab in administrator. This is not working. I tried recompiling the Sequencer Job, I also restarted DS services but it does not work.

After checking option "Automatically handle activities that fail" do I have to do anything else?

Thanks in advance
-Sumit

[chulett]

In what way is it not working? What is it not doing that you are expecting it to do?

This option doesn't seem to do a whole heck of alot, from what I read about it. And for it to work at all you have to not use your own "error" triggers in the Sequencer. Other than that, it is simply a matter of checking the box and it will add some code for you when you compile the job, all behind the scenes.

[kduke]

You need to also check add checkpoints so sequence is restartable. You need an exception handler stage. We usually route this to an email notification and a routine activity which calls UtilityAbortToLog.

[sumitgulati]

If a Server Job aborts, the Sequencer that called this server job still is shown in FINISHED status. Ideally the Sequencer jobs should also abort. I guess thats what "Automatically handle activities that fail" option is meant to do. But it does not seem to be working.

Regards,
Sumit

[kduke]

That is why you have to run UtilityAbortToLog to show this sequence as failed or aborted.

The good part of this is to automatically start where it failed. It works great.

[sumitgulati]

Kim, If I still have to explicitly use UtilityAbortToLog/UtilityWarnToLog then what is "Automatically handle activities that fail" option for?
My understanding says that if we check "Automatically handle activities that fail" in the administrator the while creating the sequencer it automatically puts an extra piece of code in sequencer job. This extra code is to reflect child job's abort in the parent job. Please correct me if I am wrong.

Regards,
Sumit

[kduke]

Not really. The exception handler traps any event that you do not trap yourself. So if you trap executed ok and warnings then it will trap any other return value. If you only have an OK link then warnings will go thru the exception handler.

The UtilityAbortToLog is so this sequence aborts so the parent sequence will see the abort.

I like the way it works. Lots of control and very flexible.

[Xanadu]

Kim, If I still have to explicitly use UtilityAbortToLog/UtilityWarnToLog then what is "Automatically handle activities that fail" option for? My understanding says that if we check "Automatically handle activities that fail" in the administrator the while creating the sequencer it automatically puts an extra piece of code in sequencer job. This extra code is to reflect child job's abort in the parent job. Please correct me if I am wrong.

Regards,
Sumit

That was exactly my thought Sumit. Whenever I am using a routine stage in the sequence my sequence fails because the routine returns a non-zero value.
(I am working on the PSFT EPM jobs - this problem comes up when doing the incremental loads..)
Did you figure out an easy way to solve the problem ?

Thanks in adv
-Xanadu

[kduke]

I think the same issue is in effect. A routine needs to return a 0 for "OK". The way a sequence works is very powerful you can tell it to ignore the result of this routine by doing an otherwise. You can also tell it to not make this step restartable which means it will not care if this step fails. Very cool. You can trap for a error an not execute the error handler. I do not see a problem here. You just have to tell it exactly what to do. It cannot guess what is an error or guess what to do. You have to trap each return code and what you do not have a trigger on is going to the error handler.

When you get used to this then you will like it. It just is not intuitive at first. It may mean you have 2 triggers on one job like "OK" and another for warnings. This means you will need a sequence with "Any" selected because you cannot get both results from one job.

Adjust to it. They did sequences right this time.

[Xanadu]

thats cool... let me restate so that i can confirm if I am understanding it right..

1) so if I have this in the sequence
Routine-->Job
2) I put the trigger in the routine as otherwise
3) enable the "automatically handle the activities that fail.." in the sequence.

Then the sequence would ignore the return code of the routine...(doesn't care about the status of the routine..)
But when the job in the sequence fails, the sequence fails too ?

Thanks Duke.
-Xanadu

[kiran_kom]

thats cool... let me restate so that i can confirm if I am understanding it right..

1) so if I have this in the sequence
Routine-->Job
2) I put the trigger in the routine as otherwise
3) enable the "automatically handle the activities that fail.." in the sequence.

Then the sequence would ignore the return code of the routine...(doesn't care about the status of the routine..)
But when the job in the sequence fails, the sequence fails too ?

Thanks Duke.
-Xanadu

Yes it would. Look at the code generated under the job control page. It will be obvould how the sequence is going to behave..

[shaimil]

For routines to work correctly I see that 0 is required for a success. What is the case for User Status. How can I use this with restartable sequences?