[ https://issues.apache.org/jira/browse/HBASE-10975?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Adela Maznikar updated HBASE-10975: ----------------------------------- Description: HtableMultiplexer is creating manually HTableFlushWorker thread for each regionserver. Changing this to use ScheduledExecutorService. Also this will help us get rid of the Thread.sleep() in HTableFlushWorker.run() and make the code smaller and easier to read. Add a unit test which will spawn 10 client threads which will do 10k puts each concurrently and check that when we do get() we get the data back. was: HtableMultiplexer is creating manually HTableFlushWorker thread for each regionserver. Changing this to use ScheduledExecutorService. Also this will help us get rid of the sleep methods in HTableFlushWorker.run() and make the code smaller and easier to read. Add a unit test which will spawn 10 client threads which will do 10k puts each concurrently and check that when we do get() we get the data back. > [89-fb] HTableMultplexer to use executor > ---------------------------------------- > > Key: HBASE-10975 > URL: https://issues.apache.org/jira/browse/HBASE-10975 > Project: HBase > Issue Type: Improvement > Affects Versions: 0.89-fb > Reporter: Adela Maznikar > Assignee: Adela Maznikar > Priority: Minor > Fix For: 0.89-fb > > > HtableMultiplexer is creating manually HTableFlushWorker thread for each > regionserver. Changing this to use ScheduledExecutorService. Also this will > help us get rid of the Thread.sleep() in HTableFlushWorker.run() and make > the code smaller and easier to read. > Add a unit test which will spawn 10 client threads which will do 10k puts > each concurrently and check that when we do get() we get the data back. -- This message was sent by Atlassian JIRA (v6.2#6252)