[FLUME-3059] Hive sink failed in hive 2.1.0

• Type: Bug
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Affects Version/s: 1.7.0
• Component/s: Sinks+Sources
• Assignee: Unassigned
• Reporter: Jinjiang Ling
• Created: 23/Feb/17 11:54
• Updated: 23/Feb/17 11:55

Description

As I want to use hive sink to load data into hive 2.1.0, then I configure it with the example in flume user guide. But something wrong happened, the error message is here:

2017-02-23 18:49:09,079 ERROR org.apache.flume.SinkRunner: Unable to deliver event. Exception follows.
org.apache.flume.EventDeliveryException: org.apache.flume.sink.hive.HiveWriter$ConnectException: Failed connecting to EndPoint

{metaStoreUri='thrift://127.0.0.1:9083', database='logsdb', table='weblogs', partitionVals=[asia, test, 17-02-23-18-40] }

at org.apache.flume.sink.hive.HiveSink.process(HiveSink.java:268)
at org.apache.flume.sink.DefaultSinkProcessor.process(DefaultSinkProcessor.java:68)
at org.apache.flume.SinkRunner$PollingRunner.run(SinkRunner.java:147)
at java.lang.Thread.run(Thread.java:745)
Caused by: org.apache.flume.sink.hive.HiveWriter$ConnectException: Failed connecting to EndPoint

{metaStoreUri='thrift://127.0.0.1:9083', database='logsdb', table='weblogs', partitionVals=[asia, test, 17-02-23-18-40] }

at org.apache.flume.sink.hive.HiveWriter.<init>(HiveWriter.java:99)
at org.apache.flume.sink.hive.HiveSink.getOrCreateWriter(HiveSink.java:344)
at org.apache.flume.sink.hive.HiveSink.drainOneBatch(HiveSink.java:296)
at org.apache.flume.sink.hive.HiveSink.process(HiveSink.java:254)
... 3 more
Caused by: org.apache.flume.sink.hive.HiveWriter$ConnectException: Failed connecting to EndPoint

{metaStoreUri='thrift://127.0.0.1:9083', database='logsdb', table='weblogs', partitionVals=[asia, test, 17-02-23-18-40] }

at org.apache.flume.sink.hive.HiveWriter.newConnection(HiveWriter.java:380)
at org.apache.flume.sink.hive.HiveWriter.<init>(HiveWriter.java:86)
... 6 more
Caused by: org.apache.hive.hcatalog.streaming.InvalidTable: Invalid table db:logsdb, table:weblogs: 'transactional' property is not set on Table
at org.apache.hive.hcatalog.streaming.HiveEndPoint$ConnectionImpl.checkEndPoint(HiveEndPoint.java:340)
at org.apache.hive.hcatalog.streaming.HiveEndPoint$ConnectionImpl.<init>(HiveEndPoint.java:312)
at org.apache.hive.hcatalog.streaming.HiveEndPoint$ConnectionImpl.<init>(HiveEndPoint.java:278)
at org.apache.hive.hcatalog.streaming.HiveEndPoint.newConnectionImpl(HiveEndPoint.java:215)
at org.apache.hive.hcatalog.streaming.HiveEndPoint.newConnection(HiveEndPoint.java:192)
at org.apache.hive.hcatalog.streaming.HiveEndPoint.newConnection(HiveEndPoint.java:105)
at org.apache.flume.sink.hive.HiveWriter$8.call(HiveWriter.java:376)
at org.apache.flume.sink.hive.HiveWriter$8.call(HiveWriter.java:373)
at org.apache.flume.sink.hive.HiveWriter$11.call(HiveWriter.java:425)
at java.util.concurrent.FutureTask.run(FutureTask.java:262)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)

It's becasue that https://cwiki.apache.org/confluence/display/Hive/Hive+Transactions#HiveTransactions-TableProperties and https://issues.apache.org/jira/browse/HIVE-11716. So since HIVE-11716 the "transactional" must be "true" when creating table like this:

create table weblogs ( id int , msg string )
partitioned by (continent string, country string, time string)
clustered by (id) into 5 buckets
stored as orc TBLPROPERTIES ("transactional"="true");

And some configuration must be setted in hive to support transaction. But the example in flume user guide of hive sink does not mention these. I think it's nessecory to specify the hive transaction configuration in the guide;