DSXchange

Version 8.1 system requirements, AIX compilers: - May 2008 XL C/C++ Enterprise Edition V8.0 for AIX PTF, Reference #4019338 - August 2008 XL C/C++ Enterprise Edition V9.0 for AIX, Reference #4020144 - October 2008 XL C/C++ Enterprise Edition V10.1 for AIX, Reference #4021392 http://www-01.ibm.com/su...

That would be the Wait For File stage I mentioned on the first reply... It's built into DataStage, so there's no need for an external product to do the same thing.

I had to edit that last post to get the quotes working right...

Anyhow, I think it's not well understood excatly what goes on and why, and not very well documented.

it is well undestood ... Not really. Yes, reference data coming into the Lookup stage should fit into memory. However, it will first build a memory mapped file, on disk. Look for dynLUT* files on your disk. Then it will load the file into memory. What happened in your case is you didn't have enough...

Search this site on "ProcessorErrorException" and find nothing but this topic! That's not good...

Search this site on "received signal SIGABRT" and you will find many topics, some of which are resolved. Maybe one of those will be related to yours.

First go into your job properties and add the environment variable APT_DISABLE_COMBINATION then run your job with that set to True. It will give the stage name having the fatal message. I would guess you got new and unexpected input data today.

Scheduling the job 24 hours in a day is an option but it will degrade the os performance and datastage performance so suggest some optimal way. Try testing that option to find out exactly how much it degrades performance. I think it would be difficult to measure. Then load test it. How many such jo...

whenever file arrive i have to run the Job. In that case, you're not likely to need to funnel two files in the same job at the same time. Schedule a sequence job and use the Wait For File stage. It's not going to pose a significant burden on the OS or DataStage performance. If you design it differe...

Total sort... Topological sort... http://en.wikipedia.org/wiki/Torrent_Systems :idea: http://en.wikipedia.org/wiki/Timsort Either tsort or psort can be used to sort a partition, so I was wondering how the T and P designators actually came about. The docs confuse the matter further by associating pso...

Version 8.1 clients should still be supported on certain editions of Windows 7 (not Home edition). So if it's causing a problem, contact Support about it. I really like viewing job logs from within Designer (Ctrl+L). That feature was introduced in version 8.5 fix pack 1, and likewise is in version 8...

See this topic regarding client/server version mismatch:

viewtopic.php?t=145302

Try the following: -username or -user instead of -u -password or -pass instead of -p -file instead of -pkg Note: "The specified file name must be relative to the deploy directory in the target system." It implies an absolute path may not work... remove the double quotes The Admin Guide sho...

I think the example is clear now. I don't see any pivoting going on. I see more of just rearranging values across columns. Try Ray's If Then Else suggestion before making it too complicated. If reading from a file, you could perhaps read all columns as 1 char or varchar column (if not then concat al...

I think it's a fine question. Considering that the Join stage requires sorted inputs, I don't think you would need a Sort stage following Join stage to begin with, unless you need to resort based on some other key.

It looks rather random to me. Can you explain what's happening between the input and desired output?