Page 1 of 1
Named Pipe
Posted: Tue May 06, 2008 12:15 am
by nshashi.85
I have a fastexport job.
when we are using datafile the following error was coming.
"Fatal: Cannot Open Datafile".
Then we ran the job using NamedPipe, which is successful.
what is the difference between the named pipe and data file?
any help?
Posted: Tue May 06, 2008 7:48 am
by chulett
You can google up the definition of a 'named pipe' if that concept is unfamiliar to you. However, that shouldn't be the difference between success and failure here.
How about explaining your job design? Saying you have a "fastexport job" doesn't really tell us much of anything.
Posted: Tue May 06, 2008 3:48 pm
by ray.wurlod
Au contraire, mon ami...
"FastExport" indicates extraction from Teradata. It may (for example) be that the user ID for Teradata lacks permission to create a file in the indicated directory but can create a fifo file (named pipe), presumably in a different directory.
Posted: Tue May 06, 2008 3:55 pm
by chulett
ray.wurlod wrote:"FastExport" indicates extraction from Teradata.
Well, ok... doesn't tell us much other than that.
![Wink :wink:](./images/smilies/icon_wink.gif)
Posted: Tue May 06, 2008 3:58 pm
by ray.wurlod
Is there any more information in the error entry in the DataStage job log? Is there any more information logged in the Teradata system?
Posted: Fri Aug 15, 2008 12:16 am
by toshea
With the named pipe option, the FastExport process will stream data to a named pipe, and the Teradata MultiLoad plug-in will concurrently read the data from that named pipe.
If you turn off the named pipe option, the MultiLoad plug-in will first run FastExport to dump the output to a sequential data file. Then it will read that sequential data file and send the data down the output link. This is slower than using a named pipe since you don't get the concurrency, but it is a workaround if you have named pipe issues. It is generally preferred to use named pipes if possible.