IBM Mainframe Forum

Long time lurker, figured I'd join, not totally unrelated to something I've yet to figure out.

For all intents, subject line says it all. Never done, never had a need.
To that end, I've started to rtfm.... IBM Network file system guide and reference.
Maybe this is not the correct manual? Seems like it is, but I do not really know.

My task is to take a mainframe sequential dataset, 133 byte, 150K record file, and place it on a server.
Dataset contains batch job statistics.

One of our Unix sys-admin says to create a NFS of the mainframe data.
Seems like a NFS is different than a normal FTP of a dataset to a server file.
Once received, his side can manipulate to make it easily available to the corporate world for 1st level investigation.

The process would be repeated every X minutes by our scheduling package.
Hopefully, this would reduce the number of requests for batch job status to OPS.

Am I on the right path?
Any thing I should ask, but don't know to ask?


Thanks much,
don t.

Hi,

I understand that you want to transfer a mainframe dataset to unix system.
If that is your requirement, Please use the JCL for transferring files from mainframe to Unix server(Provided, required firewall ports opened)
For repeating the same process every x Minutes, run this JCL as a cyclic job in scheduler.

Please use below JCL..
//JOBCARD
//FTP EXEC PGM=FTP
//SYSPRINT DD SYSOUT=*
//OUTPUT DD SYSOUT=*,DCB=(RECFM=FB,LRECL=160,BLKSIZE=3200)
//INPUT DD DSN=TEST.JCL(FTP),DISP=SHR
/*

TEST.JCL(FTP) <-- PDF member with input
-------------
xxx.xxx.xxx.xxx <--- IP address of UNIX server
username <-- User ID of UNIX mail box
password <--- Password
put 'File name' /path/XXX/YYY/ZZZ <--- Path in UNIX server
quit

If I misunderstood your requirement, please ignore.

Thanks,
$#@<+#¥

Shakthy, your solution depends upon FTP and the original post SPECIFICALLY mentioned NFS and not FTP. Providing solutions to the wrong problem is almost as bad as providing wrong solutions.

don t, I"m not sure there's a clear direction in your post. Is the system z supposed to be the NFS client or the NFS server? If the client, you'll need a server to connect to and that server will run (most likely) cron every X minutes to grab the data file the mainframe puts on the server. If system z is the NFS server, you'll need to set up the appropriate network connections so the Unix client can access the server and get the file every X minutes. In any case, there needs to be work between the system z system programmer and the network support person and the Unix support person to work out how to set up the access and then to test it.

Hello Mr. Sample.

1 - Our scheduling package is housed on the mainframe, controls mainframe, distributed, As/400 batch submissions.
Data related to run times, CC's, job stream name, etc., also housed on the mainframe within the scheduling package data bases.
1a- 90+% of our batch is distributed.
1b- Dev & support teams could log on to the server of execution, in a known location, for their status but they prefer to contact OPS/Scheduling.
We provide them with a clear/concise view vs. the granular info on the server.
1c - It is not uncommon to have 300+ such requests in a 24 hour period.
1d- We would reply with a copy/paste of the data, exactly formatted, as what I am trying to automate.
1e- Our scheduling package controls over 1.7 million batch submissions per month on aprox. 125 different servers of execution.

2 - My concept..... take the scheduling package data base records of what it knows, populate a PS dataset, send this data to "a file on a box".
First level batch status requests for Unix and windows would be directed to this "file", reducing 1c.
2a- Initial thought was to FTP the data to a box, contact someone to assist in making it a text web site. Inform our requestors to click a link for 1st level information.
2b- I contacted an in house Unix sys-Admin who says to use a NFS vs. a link. He says a NFS could be made available to the global corporate world no
matter what box the user connects. He says it's feasible to do.
2b- Any mainframe in house employee who would possibly know if we currently perform this process are NLA.
2c- I have zero knowledge of NFS, Unix.

3 - I googled NFS, scanned our production datasets for NFS statements. I did find something related to PATH= but please refer to 2c & 2b.
3a- The production code associated to the PATH= findings is created dynamically, and not all parts are readily available. Refer to 2b.
3b- I could probably reconnect the dots of 3a, but thought it might be easier to start from scratch, knowing I might have some code to refer.

Sorry for the book.
HTH, Thanks much,
....don t. .....

I'm guessing that the NFS file you refer to is within USS (Unix System Services) on the mainframe? So it's not going off platform, you just want to transfer it to the Unix part of the mainframe where others will be able to access it?

If so, you can use program BPXBATCH for this. Try Googling that and see what you get. I don't have a sample immediately to hand unfortunately. Many of the Unix parameters that can be used are documented in the JCL manual, e.g.; PATH, PATHMODE, etc.

You are going to need to work closely with your site support group. Depending on what you want, changes may be needed to BPXPRMxx in SYS1.PARMLIB. If you have not yet started reading SC23-6883 z/OS Network File System Guide and Reference you should start reading it.

Blackthorn: NFS allows access to file systems NOT under z/OS nor UNIX System Services.

Blackthorn/ Mr. Sample.....Thank you for the info.

Yes... I've started to read SC23-6883-00
I'll need more caffeine (grin).

Have fun.....don t. .....