DSXchange

Your two options are:

(a) login under eservice or

(b) send an email with the description and your license number to WDISupport@US.IBM.COM

Then you have a bad installation or insufficient access rights as mentioned in an earlier response. I still find it odd that you have no "jobs" category, but that could be explained by the access rights issue to even read the repository.

shankar_ramanath wrote:Is it possible to set the warning limit using a before-job rotuine call? Or from anywhere within the sequencer?

No, unfortunately not. You can use the command line invocation to change the warning level with "dsjob -run -warn {n}", though.

saraswathi wrote:... Call to malloc() failed to create BOR jump table
/usr/lib/dld.sl: Not enough space

malloc() is a UNIX routine that is used to get program working memory. If this call fails it means the process has no more virtual memory or that the system has no more virtual memory available.

What error message do you get? My first guess is that "dtexec" is not in your command path.

Can you resize the file to a static hashed file or at least try a "RESIZE {filename} * * *" and see if that makes a difference?

Enter the text "\jobs\testjob" in the item name box.

Actually, in this case just do a Sequential Stage -> Sequential Stage. If your delimited file has VarChar fields you will have to use a modify or transform stage to convert the datatypes to fixed width ones.

Remove the APT_ORA parameters from your job, most likely the timed commit is taking precedence.

The odd thing is that you have no "jobs" category to save into. Can you try to enter "\Jobs\testjob" to create a new folder and save into that?

You posted this in the PX forum, but marked it as a Server job ... which is it?

What method are you using to write to oracle? The commit frequency of 0 in the upsert commit interval will force the whole job to be one transaction and will not commit any records if it aborts.

What method are you using to write to oracle? The commit frequency of 0 in the upsert commit interval will force the whole job to be one transaction and will not commit any records if it aborts.

The point is that the parallel API does nothing else but call the internal BASIC routines to get at the metadata, and since the original poster already has these coded there is no reason to re-code it into c++

Please post what you tried in DataStage. You can concatenate the columns and then use the StringToDate() function.