02-05-2018 08:38 PM - edited 02-05-2018 08:40 PM
I'll give little background about this. We are inserting data into hive tables using flume through Kite Sink. This is not streaming job. Flume is started using more memory than norma and suddenly we observed the below message in the flume logs.
sinkRunner -pollingRunner-DefaultSinkProcessor org.apache.hadoop.hive.metastore.HiveMetaStoreClient.Open
org.apache.thrift.transport.TTransportException:java.net.SocketException: Connection reset
Please see more error message in below screen shot.
Why flume is connecting to hive metastore while inserting data into hive tables through kite sink?
Note: This is not streaming data.