Sorry to bother. For my MapReduce framework, I split a scan into several sub-scan of one region, however, I still got many rack-only MAP tasks rather than data-local MAP. Is this the problem of mapred.max.split.size setting because of hdfs defaul 64 mb block size? Or I need to use some other taskScheduler, i.e. FairScheduler? Many thanks.
-- View this message in context: http://apache-hbase.679495.n3.nabble.com/HBase-MapReduce-rack-local-MAP-tasks-tp4078035.html Sent from the HBase Developer mailing list archive at Nabble.com.
