Details
-
Bug
-
Status: Closed
-
Blocker
-
Resolution: Fixed
-
1.20.0
-
None
Description
With a connection to Phoenix Query Server using either storage-jdbc or storage-phoenix, two queries fail after an infinite planning loop. One query is against the Phoenix QS (c.f. phoenix-table-* attachments), the other does not involve Phoenix at all and queries Parquet in HDFS (c.f. profiles-query-* attachments). Both queries go through Drill views, the definitions of which are attached to this issue. They are both only projections.
Software versions in the environment where the bug exists: Hadoop 2, Phoenix 4.15.0 with hbase 1.5.0 and phoenix-queryserver 1.0.0. Downgrading Drill's phoenix-queryserver-client jar from 6.0.0 to 1.0.0 to accommodate this PQS version does not remediate the problem.
Storage-jdbc config.
{ "type": "jdbc", "driver": "org.apache.phoenix.queryserver.client.Driver", "url": "jdbc:phoenix:thin:url=http://localhost:8765;serialization=PROTOBUF;authentication=SPNEGO;principal=drill/bit@FOO.CLUSTER;keytab=/etc/hadoop/conf/drill.keytab", "writerBatchSize": 10000, "enabled": true }
The same storage-jdbc config is deployed in Drill 1.16 environments which do not exhibit this infinite planning bug.