Error writing to pipe: Interrupted system call

kausmone · Post by **kausmone** » Thu Apr 10, 2008 12:53 am

Hello,

I have a server job performing a Teradata Multiload using TDMLoad stage. Source data is in a flat file. The job has been running fine for months, scheduled at a frequencey of every 10 minutes. Recently however, every once in a while it aborts with the below message

Code: Select all

Could not write row to data file in stage <jobname>.<mload stagename>
Error writing to pipe: Interrupted system call

I tried re-running the job with the same source data on the advise of a colleague and it worked on every occasion. Today the job again aborted, and a re-run failed with the same error. I recompiled the job and it executed fine, and has been running fine once again since then, but this trial-and-error approach bothers...

I tried searching the forum if anyone has faced a similar problem but didnt find any related posts. Can someone help in understanding what the problem is please, and whether there really is no solution other than the no-brainer compile+rerun?

Thanks,
kaus

kausmone · Post by **kausmone** » Fri Apr 11, 2008 1:21 am

Another job in the same project doing a similar MLOAD aborted today with the same reason. And it executed successfully on a re-run. Has anyone encountered similar behavior before?

hamzaqk · Post by **hamzaqk** » Fri Apr 11, 2008 4:38 am

Did you specify the data file in the MLoad option ? , try running it with pipes option check in the general tab. should solve the problem

kausmone · Post by **kausmone** » Fri Apr 11, 2008 4:43 am

Thanks, I will try that.. but what exactly was the problem?

hamzaqk · Post by **hamzaqk** » Fri Apr 11, 2008 5:01 am

More of a OS level problem. As far as i can get it i think when you try to run it with the sequence the job aborts and then when you manually run it it works just fine ? in that case you might want to see if the user which is created when running the sequence has rights to write to a file on the system. Because in the method you selected a temp data file will be created in which all the data will be extracted before loading it into TD.
Apart from this you might also want to check the buffering in your job. how big is the flat file ?

kausmone · Post by **kausmone** » Fri Apr 11, 2008 5:08 am

The file's not too big, a few hundred records, about 500 kbytes. When I rerun, I actually rerun the entire process, so I am not sure if it is an access issue. And like I said, it doesn't abort all the time, only once in a while...