DSXchange

D'oh, I meant T30FILES and as Arnd mentions, problems with it being too low are typically pointed out directly in any error message.

And you're certain all it is doing is a truncate? That sound take... moments. Perhaps it is doing a transactional delete of records in the table instead.

Yes. And while there are upwards of 30,000 users here and only 6 or 7 of us, I would say it would be a worthy investment in your ETL future. However, anyone who feels like contributing can take a shot at helping you out. To that end... It's generally best to answer any questions asked rather than s...

Don't really see how this can be a 'sometimes' thing, unless you mean you only use the Lookup stage sometimes or do lookups with this particular config file sometimes. Or are you saying that your problematic job sometimes works correctly and sometimes generates this error?

When the job took 1 hour 13 minutes, how much of that time was spent running the before script? You can tell from the detail in the logs.

Don't blame the sequential file reads for that. 'Plain inserts' are not speedy and it can also be affected by the indexes on your target table, extents, system load, etc etc.

Sreenivasulu wrote:You need to source your dsenv by executing . ./dsenv

I'm always curious why someone would feel the need to come in later and post the exact same advice as someone else.

What is your 8.x version?

Sounds like a Sort stage is in order, perhaps sorting descending. Do you actually have to assign a 'rank' as well or simply sort them in ranking order?

Change the metadata to VarChar.

In other words, you cannot 'ignore' the missing ones like that. How about a simple 'cat' instead using wildcards? It won't care.

So that seems like that should simplify things somewhat. I'm not exactly sure how, exactly, off the top of my head. Holiday today in the States so need to head out, hopefully someone else will step in here with their thoughts.

OK... another clarification. So, it's not one or some combination of output directories, it's either one specific one or all of them? Meaning you may be able to trigger off a single output directory or something else to signify 'all' in your hashed lookup.

Obligatory message re: Windows XP not being a supported Server platform.

I'd go this route as suggested by 'oerr': $ oerr ORA 12571 12571, 00000, "TNS:packet writer failure" // *Cause: An error occurred during a data send. // *Action: Not normally visible to the user. For further details, turn // on tracing and reexecute the operation. If error persists, contac...