It’s inevitable - you’re going to encounter job failures when working with Azure Data Lake Analytics U-SQL jobs.

Error Sources

There are at least three root causes for a U-SQL job failure:

1. A problem with your script

2. A problem with the data

3. A problem with your extractor’s “shematizing” of your data

Deciphering Error Messages

Whenever an error is reported, it can be helpful to slow down and read through the feedback that the Azure Data Lake Analytics service gave you.

Here’s my general approach to deciphering error messages produced by the Azure Data Lake Analytics service:

1. Read the “Message” portion of the error to get a general feel for what the service didn’t like.

2. Decide where the root cause was: The script? The data? The extrator’s schematizing of your data?

3. Read the “Detail” section and look for the position of the ### symbols

This is your biggest clue about where the job service encountered a problem that it didn’t know how to recover from.

comments powered by Disqus