DSXchange

Then try Ray's approach to see if they are getting rejected for some reason.

You'll need to find an alternate method of delivering that first row. One possibility - don't use the automatic method but first send a static row of data to the target with those values.

GJ_Stage wrote:Tony , You are correct. There is no other reason to make it slow.

Sure there is. Do they have an index over the update key fields in the target?

High level... first you need a static source. After that it is a matter of marking your progress in the job, typically at each commit point, so you know the last successful one. That 'marker' row count gets set to zero at the end of a successful run. Each time the job runs, the marker is passed in a...

John Daniel wrote:... as Email Body !!

Just as an FYI, that alone forces you down the routine path. Without it, you could leverage the Email Notification stage in a Sequence job.

jwiles wrote:I know it's something that has been discussed many many times (I can't emphasize "many" too much) here in the forum....

We tend to say ad nauseam.

Keep in mind the fact that an 'upsert' error is posted only if the second action fails, so that's where it comes from. Any idea why your update DML might be creating duplicates? We can't know as we aren't privy to your constraint's structure nor what key fields you are using to control things.

A Sequence job with the Wait For File stage could do a simple existence check and (if found) trigger a job / routine / whatever to create your 'ok' file.

... and yet I still have this nagging remembrance of going through all this some time ago and that not being quite correct. I don't have any kind of DataStage access, so can't check the generated source code or play any reindeer games to help out so may just leave it at that, unless I am successful ...

Other way 'round, actually. 0 is false and (technically) any non-zero value evaluates as true, including the 1 used here.

My "pick up where it left off" comment was specifically directed to ETL jobs, not at the job control level. It may not be typical but it can certainly be done.

Are you saying your metadata is one varchar(500) column throughout the job and yet you still lose the header & footer? That doesn't make sense to me.

Restartability of an ETL job can also mean it picks up where it left off. And it's a workflow in Informatica, not a workload. The other difference is a workflow is required to run a single mapping, a sequence job isn't.

Except we've been through this before and from what I recall they run in a serial fashion if all you do is dump them on the canvas unconnected. I'll see if I can dig up the rather lengthy discussion...

Doesn't even need to be that 'complex', just read and send it as a varchar with no quotes and all of the records should be transferred intact.