MVSFORUMS.com

Phantom · Posted: Thu May 15, 2003 10:49 pm Post subject:

Well, is there any way to get around this problem ?

Thanks,

semigeezer · Posted: Fri May 16, 2003 12:38 am Post subject:

OK. I'm confused... that is to say, I can't reproduce this with a simple testcase.
If my TSO session has PROF INT WTP and I submit

       //USERIDA JOB ...
       //BATCH    EXEC PGM=IKJEFT01,DYNAMNBR=128
       //SYSTSPRT DD SYSOUT=*
       //SYSTSIN  DD *
       SUB 'HLQ.CNTL(IEBGENER)'
       /*

I don't get a message at my terminal that the 2nd job was submitted. I just get the Notify message when the second job ends. What am I missing?

Assuming I am missing something major, which would be no big surprise, you could just avoid the SUBMIT command altogether and just write the JCL directly to the 'internal reader'. Basic JCL for that is:

       //* SEND AN INLINE JOB TO THE INTERNAL READER
       //GENER1   EXEC PGM=IEBGENER
       //SYSPRINT DD  SYSOUT=*
       //SYSUT1   DD  DATA,DLM=99
       //USERIDB JOB ...
       //IEFBR14  EXEC PGM=IEFBR14
       99
       //SYSUT2   DD  SYSOUT=(A,INTRDR)
       //SYSIN    DD  DUMMY

but you can do it from a program by allocating the internal reader to a ddname and writing to that.

Phantom · Posted: Fri May 16, 2003 1:17 am Post subject:

Semigeezer,

I feel very sorry to say this, but really speaking I don't understand what you have suggested. Let me try to explain my situation.

I already have a big JCL which requires a file which is available only in PRODPLEX. One way is to manually NDM the file from PRODPLEX and run the JCL (JCL should run only in TESTPLEX). But, I wanted to automate this stuff.

I had another JCL which NDMs the file from PROD to TEST but that JCL can run only in PRODPLEX. Instead I used JOBPARM SYSAFF=333C and tried to submit the JCL from test. It works fine.

Now, I need to put everything together. As far as I know, JOBPARM SYSAFF cannot be specified for a particular step alone. So I wrote a REXX program which inturn submits the NDM Jcl and also wait untils the file gets copied to testplex. Then this file is used in the subsequent steps of the first JCL as input.
This is the scenario. I wanted to the whole process without any user interaction until the job completes. But Once the NDM jcl is submitted I receive a submitted message and also another message saying that the JCL is transferred to PRODPLEX. I want to stop this messages from being displayed.

I haven't yet tried ur suggestion. Still, If you submit a JCL from another JCL, will the first jcl wait until the submitted (second) JCL completes ??? This is my question. Because the next step in JCL1 requires the input file, But it takes some finite time for the file to be transferred from PROD to TEST even if it is a small one.

Please correct me if I'm wrong at any point. If you have a better solution for my problem, please advice me. It would be of real help.

Mike · Posted: Sun May 18, 2003 7:13 pm Post subject:

Phantom,
no the first job will not wait for the second to complete unless you code this within a program. If the submit message is a problem as well then I'd go for some code along the lines of :-

Phantom · Posted: Sun May 18, 2003 11:26 pm Post subject:

Mike,

Thanks for your suggestions. I see that, semigeezer had given me a similiar or infact same solution but through JCL. I would try your suggestions and let you know. I need to get more details regarding Internal readers. I would appreciate any help from you all regarding this.

I also would like to know more about REXXWAIT. Can you please post the code. As you can find in my earlier posts, I had used

Mike · Posted: Mon May 19, 2003 9:09 pm Post subject:

Phantom,

regarding internal readers there's nothing that special about them, they are, as far as I'm aware, just a special writer (printer) that instead of printing the ouput passes the output to JES. Basically when you use the SUBMIT command it wil be passing the data to the internal reader. What you might have to check is whether your site standards allow their use (probably not for production jobs).

Regarding the SYSCALL command if it works then it's probably no worse or better then the REXXWAIT routine, however you've got it installed. Basically things like this simply call the STIMER MACRO with the appropriate parameter. The one I've written just has a Rexx interface, which isn't necessary. The main thing iin using such a utility is that it avoids using a loop that will clock an impressive (or not depending upon your viewpint Smile

) amount of CPU. I'll include the code for REXXWAIT at the end as it's quite long.

Regarding the NDM PROC here it is :-

Phantom · Posted: Tue May 20, 2003 4:52 am Post subject:

Thanks Mike.

I tried your suggestion of using internal readers. I have one more question in that. Now, I'm not receiving the Job Submitted message. Still, I receive 2 messages. 1. a message that says that the Job has been transferred to the PROD region. (Since I have used JOBPARM SYSAFF=333C) and another message, once the job completes in PRODPLEX. I'm not sure whether these 2 messages can be stopped by removing the NOTIFY parameter in JOB Card. Well, I haven't yet tried it. Can it be stopped by removing the NOTIFY parameter ?

My next question is regarding the NDM process. Moderators... I start to feel that this thread is breaking into a new discussion. But still for continuity I would like it this way. I hope you wont' mind this.

Mike, as far as I know, in the NDM Process, PNODE Stands for Primary node and SNODE for secondary node. Now, whenever we NDM a file using a process similar to what you have shown, the received file gets created in Scratch. I just tried to fancy my chances by writing a NDM process in my own library with DCB Paremeter overrides. But It failed just as I had expected. Have you ever created a new NDM Process? Can we ever create one ? and If yes how ? In our shop we have a process similar to what you have given us but without the S2 step. Now If I want to use your process. How do I do it ?

Mike · Posted: Tue May 20, 2003 8:08 pm Post subject:

Phantom,
regarding the messages, I'm pretty sure that at least the second message is dependant upon the NOTIFY, no too sure about the first. But try removing the NOTIFY. You could try XEQ (e.g. /*XEQ jesnode) which is what I perefer to use.

Regarding NDM processes, yes I have created my own, in fact the one that I sent was created specifically for the work we do in our team (we create and populate testing environments for UAT testing). Basically when you invoke DMBATCH the process name being used must be a member in the dataset(s) allocated to the DDNAME DMPUBLIB.

Here's some JCL that creates a temporary PDS (CRTDS step), populates the appropriate member (MITICT) with the NDM process called MITICT (CPYPDS step) and then performs the transfer via the NDMBATCH program (STEP001 step). This example is pretty old and is run (via XEQ) on the PPRD system. from the more recent process I sent yesterday, you can see that we now use the recipient system as the PNODE. and therefore use COPY FROM(SNODE .......)

Phantom · Posted: Thu May 22, 2003 4:04 am Post subject:

Hi Mike,

Thanks for all your help. Everything is working fine now.

Regarding the NOTIFY parameter. I tried to invoke the JCL without the NOTIFY parameter, This time I received no messages (JCL Transfer message, as well as the JCL completion notification message). I get the same response even if I manually turn off INTERCOM (ie. PROFILE NOINTERCOM) and then directly run the JCL. But I don't know why this doesnot happen when I issue the PROFILE NOINTERCOM command in the REXX routine which invokes the JCL.

Now, regarding the NDM Process. I created a new NDM process, following your suggestions with all the parameters that I wanted to have and It worked absolutely fine.

Here is the NDM process that I Created.

Mike · Posted: Thu May 22, 2003 4:25 pm Post subject:

Glad to see that you're happy Phantom and glad to be of assistance.
_________________
Regards,
Mike.