DSXchange

What a singluarly cumbersome approach! For example, why not use DSGetNewestLogId() and filter for warnings? It will return 0 if there are none. Or process the dynamic array returned by DSGetLogSummary() with a REMOVE statement in a loop. That way one can quickly filter on severity for each and every...

That's not many, unless each is Megabytes in size.

Welcome aboard.

Your advice is poor. One first needs to establish that the lock is not validly held - for example someone else may be editing that job.

Arbitrarily releasing locks is "worst practice" and can lead to lost work.

You've missed the point. An OK trigger will fire ONLY if the job finished with no warnings. Zero. If you want execution to continue if there are warnings you need either an additional, Warning, trigger into the Sequencer from each Job activity, or to replace the OK trigger with a Custom trigger that...

I meant your parallel job configuration - the one specifed by APT_CONFIG_FILE. This is totally irrelevant for server jobs. How many nodes are there, and where are they?

The only way you can "do it unix" (sic) is to re-boot the server, which is overkill. You need, using DataStage tools, to identify the owner of the lock, determine whether the lock is validly held and - only if it is not validly held - unlock it. If it is validly held you should negotiate with the cu...

What is your configuration? Are the Teradata libraries accessible to all nodes? In particular, is libcliv2.so accessible (execute permission) and findable through your shared library search path?

We're not gonna take it!

Very few (none) of my clients uses XML in their ETL strategy.

Could be you're not looking in the log for the correct job.

The error messages are being generated by a Transformer stage, but there is none in your posted job design.

I would suggest that you've left some old derivations in the poorly named Transformer_165 stage.

Do you feel that that would be a popular - even acceptable - "solution" in a production environment?

I don't.

Yes, but they won't work unless the pathnames for every resource disk are identical. The pathnames of the data files are stored in the control file (myDataSet.ds), but not in a text format.

And the version of DataStage must be identical, and the parallel job configuration effectively identical.

No but, if you enable server side tracing and then do it (open Manager and press Generate), you will be able to discern what happens, and possibly then to emulate the same in your script.

Time to mark this thread as Resolved, then.

Why was APT_AUTO_TRANSPORT_BLOCK_SIZE ever set to anything but False?
Transport blocks are only relevant for transmitting fixed-length records.