[
https://issues.apache.org/jira/browse/STORM-1030?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14997910#comment-14997910
]
ASF GitHub Bot commented on STORM-1030:
---------------------------------------
Github user harshach commented on a diff in the pull request:
https://github.com/apache/storm/pull/871#discussion_r44365638
--- Diff:
external/storm-hive/src/main/java/org/apache/storm/hive/bolt/HiveBolt.java ---
@@ -87,7 +87,7 @@ public void prepare(Map conf, TopologyContext
topologyContext, OutputCollector c
}
}
this.collector = collector;
- allWriters = new HashMap<HiveEndPoint,HiveWriter>();
+ allWriters = new ConcurrentHashMap<HiveEndPoint,HiveWriter>();
--- End diff --
We are running heartbeat as separate timer thread. Every time we add a new
HiveEndPoint, HiveWriter we add this to allWriters Map and heartbeat thread
goes through all the values to call heartbeat method. To have the values
visible in both threads we need to make concurrent
> Hive Connector Fixes
> --------------------
>
> Key: STORM-1030
> URL: https://issues.apache.org/jira/browse/STORM-1030
> Project: Apache Storm
> Issue Type: Bug
> Components: storm-hive
> Reporter: Sriharsha Chintalapani
> Assignee: Sriharsha Chintalapani
> Fix For: 0.11.0
>
>
> 1. Schedule Hive transaction heartbeats outside of execute method.
> 2. Fix retiring idleWriters
> 3. Do not call flush if there is no data added to a txnbatch
> 4. Catch any exception and abort transaction.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
