I've tried using sqoop import with Avro and it works beautifully. But
I've tried it with compression which according to the documentation
should work. It didn't for me so before I go any further: Is this a
problem in the documentation?
I briefly looked at the AvroOutputFormat and it doesn't do any
compression as far as I can tell. The job.xml is certainly correct in
that it mentions mapred.output.compress=true.
Removing the --as-avrodatafile option gives me a compressed text file
so I'm pretty sure I've got the command line right.