Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
0.7.0
-
None
Description
The following query does a LIMIT inside a subquery to force a UnionExchange and then does an ORDER-BY outside that will first re-distribute the data before sorting. It results in a DecoderException in netty.
0: jdbc:drill:zk=local> alter session set `planner.slice_target` = 10; +------------+------------+ | ok | summary | +------------+------------+ | true | planner.slice_target updated. | +------------+------------+ 0: jdbc:drill:zk=local> select t2.o_custkey from (select o_orderkey, o_custkey from cp.`tpch/orders.parquet` t1 group by o_orderkey, o_custkey limit 10) t2 order by t2.o_custkey; Query failed: Query failed: Failure while running fragment., refCnt: 0, decrement: 1
Here's partial output from the logs: (will attach full error log).
io.netty.handler.codec.DecoderException: io.netty.util.IllegalReferenceCountException: refCnt: 0, decrement: 1 io.netty.handler.codec.MessageToMessageDecoder.channelRead(MessageToMessageDecoder.java:99) [netty-codec-4.0.24.Final.jar:4.0.24.Final] io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:333) [netty-transport-4.0.24.Final.jar:4.0.24.Final]
Attachments
Attachments
Issue Links
- contains
-
DRILL-1995 UnlimitedRawBatchBuffer should not enqueue incoming batches when its state is KILLED
- Resolved
- is duplicated by
-
DRILL-1995 UnlimitedRawBatchBuffer should not enqueue incoming batches when its state is KILLED
- Resolved