xy7 created SPARK-18957:
---------------------------
Summary: when WAL time out, loss data
Key: SPARK-18957
URL: https://issues.apache.org/jira/browse/SPARK-18957
Project: Spark
Issue Type: Bug
Components: Block Manager
Affects Versions: 1.6.2
Environment: hadoop2.7.1 spark1.6.2
Reporter: xy7
Priority: Critical
Fix For: 1.6.2
When sparkStreaming's blockManager writeAheadLog timeout, it cause
keepPushBlocks to be stoped, with this warn log
"WARN scheduler.ReceiverTracker: Error reported by receiver for stream 0: Error
in block pushing thread - java.util.concurrent.TimeoutException: Futures timed
out after [30 seconds]".
But the receiver continued to get data from MQ, so the data lost. In sparkUI,
we would found all the batchs turned to 0 events.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
