View previous topic :: View next topic
|
Author |
Message |
sijayapal
New User
Joined: 23 Dec 2008 Posts: 69 Location: India
|
|
|
|
Hi,
We have bunch of dataset triggered jobs. Here is how it works.
1. We have a java screen to request for a job cycle.
2. Two files get created out of which one triggers the job and the another one has some details like who requested, type of job etc.. to be updated in a table to mark beginning of the cycle.
3. The cycle continues till end and finally updates the table to mark the end of cycle
I am facing JCLERR, due to a missing file. The second file which has all the information is not getting created. I have no clue how they are creating the files from the Web screen. Any one can guess that would be helpful.
Note: I am not pro on the front end and only take care of backend... and the one who knows both is not available right now...
The complete front end guys doesn't understand it making my life difficult... |
|
Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
|
|
|
|
What would be the point of us guessing?
If you want some actual assistance, you're going to have to show the JCL and the message produced, at the very least. In the Code tags please. |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19243 Location: Inside the Matrix
|
|
|
|
Hello,
You need someone to show you how the "front end" creates these files and if there is any diagnostic information available then the file create fails.
Is this possibly done by another job submitted from java? Or does the front end merely create the file (when it works) with no mainframe job running?
Keep in mind that the missing file jcl error is probably not the culprit, but rather the victim of the other process failing.
And as Bill mentioned, we need to see any diagnostic information you have available. |
|
Back to top |
|
|
sijayapal
New User
Joined: 23 Dec 2008 Posts: 69 Location: India
|
|
|
|
I don't think mainframe job creates those files...
Can a job submitted in Java create a mainframe file?????
If yes, then that could be the issue.. |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19243 Location: Inside the Matrix
|
|
|
|
Hello,
Quote: |
I don't think mainframe job creates those files... |
Probably not. As i mentioned, the failing job is the victim.
Quote: |
Can a job submitted in Java create a mainframe file?????
If yes, then that could be the issue.. |
Well, Yes. Depending . . .
Suggest you talk with your scheduling people to see if anything is defined for the problem dataset. You also need to work with the front end people and learn just "how" the problem dataset is being created.
Many scheduling systems are set up to submit a job on the arrival of a dataset.
Your situation may be that the first dataset is created and the job is triggered. It may run so quickly, that the jcl error happens while the second dataset is being created. |
|
Back to top |
|
|
Craq Giegerich
Senior Member
Joined: 19 May 2007 Posts: 1512 Location: Virginia, USA
|
|
|
|
Why 2 datasets? |
|
Back to top |
|
|
Akatsukami
Global Moderator
Joined: 03 Oct 2009 Posts: 1787 Location: Bloomington, IL
|
|
|
|
It's the requirement. |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19243 Location: Inside the Matrix
|
|
|
|
Hello,
Maybe this is reaching too far, but if both files are "created" serially by the front end, might be a matter of using the "second" dataset to trigger the job?
If the first process failed, the second should not run. If the first process is successful, the second would run and That dataset could trigger the mainframe job
If the front end processes are not run serially, they should be . . . |
|
Back to top |
|
|
sijayapal
New User
Joined: 23 Dec 2008 Posts: 69 Location: India
|
|
|
|
The first dataset is just dummy.. it is created to trigger the job cycle.. and the second one has the information like who created, cycle type etc..
Ahhh finally front end understood that they are creating the files and not the mainframe job.. they are FTPing the file to mainframe and the trigger file is created successfully but the second one they are not sure... |
|
Back to top |
|
|
sijayapal
New User
Joined: 23 Dec 2008 Posts: 69 Location: India
|
|
|
|
Craq Giegerich:
Only the developer knows... if it was me i would have used just one.. |
|
Back to top |
|
|
Nic Clouston
Global Moderator
Joined: 10 May 2007 Posts: 2454 Location: Hampshire, UK
|
|
|
|
If you are going to continue with 2 files then set your scheduler to await BOTH files before kicking off the cycle. |
|
Back to top |
|
|
sijayapal
New User
Joined: 23 Dec 2008 Posts: 69 Location: India
|
|
|
|
Quote: |
If you are going to continue with 2 files then set your scheduler to await BOTH files before kicking off the cycle. |
That sounds good but atleast now even if it fails our support team creates the file and restart the job to kick off the cycle.. else the business end would freakout that the system is down and if the front end takes lot of time to fix it.. the its impact will be huge... |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19243 Location: Inside the Matrix
|
|
|
|
Hello,
From what you just said there is no reason for the "dummy" dataset . . .
If "they" insist on creating the dummy dataset, hopefully, you can get the mainframe schedulers to trigger the job on the arrival of the "real" data rather than the dummy data . . . |
|
Back to top |
|
|
Anuj Dhawan
Superior Member
Joined: 22 Apr 2006 Posts: 6248 Location: Mumbai, India
|
|
|
|
Quote: |
Ahhh finally front end understood that they are creating the files and not the mainframe job.. they are FTPing the file to mainframe and the trigger file is created successfully but the second one they are not sure... |
1. Ask the "front end" to FTP the second file as well? JAVA has no magic-stick to send you a data just like that; they have to use some sort of transfer-protocal to have the file on zOS box. And you need to know wht is that...
2. As your Job is "triggered by", if you use CA-7 - add the second data as "requirement". That way you won't get the abend but in the case "file is not arrived", Job will in go in "late queue" instead of JCL-err/abend. I believe, with what you've explianed it's a better situation to be in. |
|
Back to top |
|
|
sijayapal
New User
Joined: 23 Dec 2008 Posts: 69 Location: India
|
|
|
|
Now i face another issue... the front end is perfect. The job is getting triggered somehow even if none has requested for a cycle.
Since none has requested for a cycle, the 2 datasets are not created. But the job somehow gets triggered and looks for the dataset and fails.
This is happening daily. Checked the ESP and it all look good to me. Could anyone help me where else to look exactly in ESP Scheduler?
If you want i can paste what is there in the ESP... |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19243 Location: Inside the Matrix
|
|
|
|
Hello,
Sounds like the schedule is defined to start the job on dataset arrival as well as every day.
Work with the scheduling people to make sure the job is only started on dataset arrival. |
|
Back to top |
|
|
bodatrinadh
Active User
Joined: 05 Jan 2007 Posts: 101 Location: chennai (India)
|
|
|
|
Hi sijayapal,
Quote: |
Checked the ESP and it all look good to me.
|
You checked the ESP code or you did manual simulation?
Please show the details..
Thanks
-3nadh |
|
Back to top |
|
|
Nic Clouston
Global Moderator
Joined: 10 May 2007 Posts: 2454 Location: Hampshire, UK
|
|
|
|
Quote: |
Now i face another issue... |
So, after over 4 years on the forum you should know to start a new topic. |
|
Back to top |
|
|
sijayapal
New User
Joined: 23 Dec 2008 Posts: 69 Location: India
|
|
|
|
Dick scherrer - Yes i had contacted.. i am waiting for a response...
Hi bodatrinadh
I checked the ESP code. I saw the triggering dataset name and the other job dependencies.
Hi Nic Clouston,
You are right but having multiple thread for same issue is not a good idea... so i just continued on the same... |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19243 Location: Inside the Matrix
|
|
|
|
Hello,
Quote: |
You are right but having multiple thread for same issue is not a good idea... so i just continued on the same... |
If this is the very worst thing that happens on the forum this year, it will (indeed) be a Very good year
d |
|
Back to top |
|
|
bodatrinadh
Active User
Joined: 05 Jan 2007 Posts: 101 Location: chennai (India)
|
|
|
|
Sijayapal,
Job dependencies are normally viewed in Consolidated Status Facility (CSF) or Event Simulation.
May be your jobs are triggered by different application or Event...
At my site, if any job triggers from ESP it will display - Submitted Time date day, System <DEVL\UAT\PROD>, Subsystem <DEVL\UAT\PROD>
Requested by event <event name>, Event triggered by user, JCL Copied to <PDS dataset name>, JCL triggered from <PDS dataset name>, Job is in application <appl name> , generation id, Job tag <tag name>
Ex:-
Code: |
//* SUBMITTED BY ESP AT 16.33.45 ON WEDNESDAY JANUARY 23RD, 2013
//* SYSTEM ESPP, SUBSYSTEM ESPP
//* REQUESTED BY EVENT PRODXXXX.XXXXX12_XXX211X9
//* EVENT TRIGGERED BY USER D416730
//* JCL COPIED TO XXX.XXX.PROD.XXXXXX
//* JCL FROM XXX.XXXXXX.XXX.XXXXX.XXXXXX(XXXXXXX)
//* JOB IS IN APPLICATION XXXXXXX, GENERATION 126
//* JOB TAG XXX_C_CYB2119
//* |
By taking the Event name we can trace where the jobs are triggered from...
Thanks
-3nadh |
|
Back to top |
|
|
|