MVSFORUMS.com

[jalves]

Hi,

i'm a beginner in JCL but i have a huge problem to solv.
I need to export files from IBM to Unix to load a datawharehouse. Currently i use a step like this:
//**********************************************************************
//* FTP DO FICHEIRO GFTPAGA0 *
//**********************************************************************
//FTPGPS1 EXEC PGM=FTP,REGION=4096K,COND=(4,LT)
//*STEPLIB DD DSN=TCPIP.SEZALINK,DISP=SHR
//SYSTCPD DD DSN=TCPIPP.CONFIG(TCPDATAP),DISP=SHR
//SYSFTPD DD DSN=TCPIPP.CONFIG(FTPDATAP),DISP=SHR
//INPUT DD *
1.2.3.4
xpto
otpx
cd mydir
PUT 'PSIG.DW.SPTCACT0.DWHOUSE.TEMP' SPTCACT.dat
close
end
//OUTFILE DD SYSOUT=*
//SYSPRINT DD SYSOUT=*
/*

What i would like to do is change the file name of the destination file by adding a numeric value stored
in a DB2 table or in a dataset. This value would be increased and updated each time the job run.
In the previous job that would be replacing de "SPTCACT.dat" name with "SPTCACT001.dat" at runtime.
After next execution the output would be "SPTCACT002.dat" and son on.
Can any one help me ?

[superk]

Well, I don't know DB2, so can't help you there. Certainly, a process to read, increment, and store a sequence counter is a trivial exercise.

Have you consider using the FTP "StoreUnique (SU)" option? If your FTP server site supports it, maybe it would be easier to implement.

Here's a link to a description of the SUnique option.

[jalves]

I have tried the FTP with "StoreUnique (SU)" option. It does not solv my problem because, after loading
the data into the warehouse, i move the file from the destination directory to a backup directory. The SU
option needs to find the same file name in the remote server directory.
Besides i need all the file names to be unique since the first extraction.
Thanks, anyway!

[kolusu]

jalves,

You can use EZACFM1 to create a dynamic file names. Check these links

http://mvsforums.com/helpboards/viewtopic.php?t=4664&highlight=ezacfsm1

http://mvsforums.com/helpboards/viewtopic.php?t=2353&highlight=ezacfsm1

Hope this helps...

Cheers

Kolusu
_________________
Kolusu
www.linkedin.com/in/kolusu

[jalves]

kolusu,

thanks for your advice. But it did not solv my problem completely.
Based on the information provided by the links i created the following job.

[kolusu]

jalves,

there is no symbolic to increase the seq number. You need to store the file as unique. so I suggested to use the date , time as part of the filename so that they are unique

Kolusu
_________________
Kolusu
www.linkedin.com/in/kolusu

[jalves]

kolusu,

i'm not concerned about increasing the seq number any more. I can do that with a select statement or a REXX program.
What i'm asking, to you or anyone, is some help to find out why the output of the previous job does not use the
value stored in SEQ variable inside the CMD variable, or if there is another way to do it.
I understand you must have better things to do but allow me to tell you the all story.
Each time our datawarehouse extraction process executes, the unix server receives data files with the newest
information stored in the online database. Due to several problems sometimes we must execute the DW loading
processes manually. Thats when, once in a while, the order of the files is confused and our DW database
becomes inaccurate. To solve this we intend to change our programs in order to keep track of the next file number
to be loaded. Programs will not process file 3 if they are expecting file 2. Thats why we can't use date and time in
file names. Our programs need to know the name of the correct datafile to be loaded next.
Any ideas ?

[kolusu]

jalves,

Here is an idea. This will increase the seqnum everytime it ran.

Create a 80 byte FB recfm format. It will have just 2 lines to begin with.
Populate the above defined file with this

let us say this filename is userid.ftp.seq