3

I'm trying to query a whole directory in S3 containing parquet files. The query hangs for a while, then returns an error:

0: jdbc:drill:zk=local> select * from s3.`/data/dt=2015-10-15` limit 10;
Error: CONNECTION ERROR: Connection /13.14.15.254:54549 <--> /13.14.15.254:31010 (user client) closed unexpectedly.

I can see my data folder when running these commands:

USE s3.`root`
SHOW FILES

Thanks!

Michael Spector
  • 36,723
  • 6
  • 60
  • 88
  • http://stackoverflow.com/questions/31617054/apache-drill-table-not-found-on-s3-bucket/31621528#31621528 addressed a problem that you might also be having. Try S3n instead of S3 and the other suggestions. – catpaws Oct 18 '15 at 05:46
  • @catpaws The question is about a different problem, definitely not about querying directories. Also, I tried using both s3 and s3n protocols. – Michael Spector Oct 18 '15 at 06:06
  • Ah, maybe a Hadoop dependency problem: https://issues.apache.org/jira/browse/DRILL-3749. This sounds like you can see the data but might not have access: https://issues.apache.org/jira/browse/DRILL-3467 – catpaws Oct 18 '15 at 06:51

0 Answers0