I wish HDFS had async or batch operations. Could possibly use a thread pool here to speed up the case of many dirs. I think the HDFS client will batch request from multiple threads into a single RPC to namenode. This could be a follow on issue. Since multiple places in Accumulo code need to create or delete lots of files, could build these high level operations into volume manager.
[ Full content available at: https://github.com/apache/accumulo/pull/575 ] This message was relayed via gitbox.apache.org for [email protected]
