DSXchange

If all you want is this particular entry, you will need to do a little more work (a routine). Scan backwards from the newest log ID until you find this particular event. Then write it to a file to be attached, or call DSSendMail() directly from your routine.

Server jobs execute in what is effectively a virtual machine called the DataStage Run Machine. This is what you install as the "DataStage server" and is also used to execute job sequences. The DataStage Run Machine is written in a combination of C and DataStage BASIC. Parallel jobs execute something...

I apologise (not really) - my RMM stage appears to be out of order.

The problem is in the Sequential File stage; DataStage has scanned 100KB without finding the record delimiter you have specified on the format tab, and so has given up. If the source file is unterminated, then it must be fixed width, which is unlikely for XML. Therefore, I can only suggest that you ...

There's no such thing as a left inner join. Are you using a left outer join, or an inner join? An inner join will return a number of rows no greater than the smaller of the row counts of the two inputs. A left outer join will return a number of rows no greater than the number of rows in the left inp...

Are you using Entire partitioning, by any chance? Even if you're not, the ODBC Enterprise stage emulates parallelism by creating as many connections to the data source as there are partitions for it to run in. This explains the duplicates; you are running the same query on each connection. Change th...

What is "proper" sorting? It's not useful to tell us it is "not proper" - we can't help you to solve that without knowing what "proper" is! And you can use any sort utility you like (for example a UNIX sort command or a DataStage Sort stage or plug-in stage, like CoSort, to get the data sorted). Use...

And your point is...?

Not necessarily. The message simply informs you that "true" floating point is not fully supported in DataStage parallel jobs; only a limited amount of precision is possible. Plenty, but limited. Use a message handler to demote this message to informational.

Multiload does set table locks, as per documentation. You could switch (within the Multiload stage) to using TPump, which only sets record level locks.

When you import a table definition from this table, what is the data type of the column in question? Did you try this data type in your job design? Does it work with VarBinary, or do you explicitly need LongVarBinary?

Use a Column Generator stage to create a key column on each Data Set. The key column has a constant value, say 1, for all rows. Partition the smaller table as Entire. Effect an inner join based on this key.

Another reason is time zones, and the need to sleep occasionally. Your specification is imprecise. Job parameters have as many as three values: the design-time defaults, created in Designer; the run-time defaults, set in Director, and the actual run-time values, set when a job run request is issued...

Export the job. Examine the export file. Is there any user-defined SQL that refers to a non-existent table, that has been abandoned in favour of generated SQL? If so, in the job design switch back to user-defined SQL, delete any SQL found there, and switch back to generated SQL.

If there is a line terminator, then you can read using ReadSeq rather than ReadBlk, or even without the routine. If you do use a routine, it would be selected as the Before Job subroutine (or Before Stage subroutine in the Transformer stage being fed by the Sequential File stage). But you can read w...