DSXchange

You do not (and cannot) know which records are in DATA.30 and OVER.30 and, indeed, they can move between the two as the dynamic hashed file rearranges itself. So please do not make the statement that the ones you want are in DATA.30. It's just not correct. The field names F1 through F10 are, as you ...

Depending on the operating system there may be a limit on the number of subdirectories that you can have in a directory. Since each DataStage job creates up to ten subdirectories, this is straightaway a potential limit on the number of jobs you could have in a project on those operating systems. Tha...

What's the exact error message? What does it tell you?

Murphy was an optimist.

Muphry's Law applies to typesetting.

This functionality is available in version 9.1, in the form of Workload Management Queues.

It's not documented. Essentially it's a hash table keyed by the combined value of the grouping keys and containing the aggregation results for that group.

Try using Array Size 1 for DELETE.

Welcome aboard. Read the file records as a single field, add line numbers, use a Difference stage to identify any difference, and join on line numbers.

Have you looked at tuning the hashed file, particularly reserving the disk space they need? Are the records very large?

Not a silly question at all. You've learned something about systematic diagnosis, as have the others who've read this thread.

As noted, CDC for DataStage can use the MQ Connector - you don't need the (old) plug-in and therefore don't need to re-install.

Look in the job log. Is the SQL actually used logged? Have quote characters been stripped?

Also, when I enable InPrivate browsing (which, among other things, blocks cookies) the problem does not occur.

Perhaps I'll try purging all Internet history including cookies.

This week whatever it is is intercepting certain key words such as Job and converting them into hyperlinks that pop up advertising sites.

It's only on my laptop; I can still DSXchange from work and from other computers.

Further investigation is clearly needed!

Check the spelling of every identifier in the Oracle Connector.