DSXchange

That is a good choice. The Oracle Migration Workbench I think is the name.

If you need to create the tables with DataStage then there is a routine to create the DDL from records in DS_METADATA called KgdCreateDdlMetadata. It is on my tips page. I think it has been posted on this site too.

My aim is to insert a row from A if it is not in B into C table.

If a row is not in B then the keys are not there otherwise explain yourself better.

You need to search for DEFFUN to call these functions. First of the month is easy. You need to oconv(@DATE, "D4-") then substring it to get year and month then put it back together. Today = oconv(@DATE, "D4-") ThisYear = field(Today, "-", 3) ThisMonth = field(...

If you are saying the keys are missing then lookup A keys into B table and if not found then insert into C. Have a second job to lookup B keys into A table.

If it is more than keys then you might be able to use MINUS queries if the tables are small and if your database has MINUS.

I am not sure how to do this in Windows. So if this is Windows then you need to do somthing similar.

If you are on UNIX then this is an easier problem to fix. In UNIX the users need to all be in the dstage group. If you run multiple groups then you need to open up these files. As root: cd `cat /.dshome` chmod -R 777 .DBsetup.log DSOStemp* DS_LICENSE UV_* catdir sql unohist chmod -R o+rx bin chmod -...

I misread the original post. These are straight table copies. I was think this was 400 jobs that need to be converted from DB2 jobs to Oracle jobs. This is much easier. Do one job exactly like you want. Then import all the tables. Export the one job and all the tables and send it to me. I can genera...

bcarlson

I am not sure I follow what you are trying to do.

logic

I think each job just takes a few minutes. Ever try to switch stage types in a DSX? If so post your code.

You need to get the iconv() of the first of the month then subtract one. Next oconv() it back.

ExecDOS and ExecSH mess up your quotes. Do it with execute. It should work fine.

Why can't you put an Oracle stage on the palette and drag the link over from DB2 to Oracle and save everything but the connection information. As long as they do not change the column names then just import the new tables and overwrite the column definitons to change the data types. I would think yo...

It would be possible to go through the log and read all the records with row counts and parse them. Then write those to a hash file. Maybe have a second job write them to a table in Oracle or your database of choice.

I doubt if anyone has done this. If they have then maybe they will share.

Sorry, long day. Fixed.

If you are getting the values from the current row then it should be easy to check for nulls and then fill in the values. If you need values from previous rows then you could do with Stage variables. Do a search. There are lots of posts using stage variables to save previous rows values.

If you are on UNIX then EtlStats will do all this. There is a parameter which is called EmailFlag. If this is not 0 then it will email this as an attachment. It uses Email.ksh to send these attachments because this did not work well in older versions of DataStage. It works great this way and I have ...