DSXchange

You've now had two people recommend not using the FTP stage. You state that the command stage is unzipping... but that the file is corrupt. You need to find out where your error is happening. Try FTPing a NON-zipped file. Does that work? If yes, then your unzip portion is introducing the problem (I ...

<sigh> As mentioned earlier, don't use DELETE in DB2, do an import from /dev/null.

I don't know TX, but usually x**y will do it (at least that will work in server/px) - have you tried that?

DB2 is a bit tricky when it comes to efficiently deleting data, usually the approach is to use a "IMPORT FROM /dev/null OF DEL REPLACE INTO TableName" command and you can put that into the DB2 stage OPEN command.

I think you might have imported the job and binary from your development machine and that is at a different rellevel than UAT. Recompile the job in UAT and consider upgrading whichever machine is at a lower version.

Robindwh wrote:...what is real significance of Look up keys if we can achieve the required output by using constraint...

You cannot achieve the same results. Lookups are used to join data to the stream. Constraints are used to direct or limit output streams.

Which stage are you using to do this? Most DB stages will support a "before" command.

Seya - please re-read what I requested.

You didn't mention that you were using the FTP DataStage stage instead of a unix/windows FTP command before. The FTP stage is best when transporting data on a row-by-row basis for processing. It is not intended for what you are doing. I suggest you use a normal FTP command to get your file, then gun...

Reset the job in the director and look at the entry labelled "from previous run..."; if the problem is still unclear please post that message contents here.

Does the job abort right away or after running a while and do you have different ulimit or uvconfig settings between dev and prod?

Constraints and keys are very different. Keys are used in lookups. Constraints are used to decide whether data moves down a particular link. The two are not related.

I do not understand; are you saying that the number of rows coming into the IPC stage and going out is not the same in a job run or are you stating that your number of rows is different between development and production?

The "Key Column" is not interpreted unless you are going to or referencing a database or Hashed file stage.

I don't know what the issue is. You have FTP'd the file from UNIX to Windows and issued gunzip to unzip the file. Can you use notepad.exe or wordpad.exe to view this file? if you do a "wc -l" on both UNIX and Windows copies are they identical? If they are identical, what happens when you try a view-...

Perhaps you could give some more information on the error. Is this an excerpt from the log? Could you post the whole error message? What sort of a stage is this error happening in?