By

Contents

create ‘UserActiveDate’, {NAME=>’lastActive’, VERSIONS=>3, MIN_VERSIONS => 1, TTL=>15552000, COMPRESSION => ‘SNAPPY’}, {NUMREGIONS => 20, SPLITALGO => ‘DecimalStringSplit’}
– 保留最新一个状态. TTL设置后即使有MIN_VERSIONS，spark查询仍然为空 ==> 数据少是没落盘文件导致 未flush

hbase org.apache.hadoop.hbase.mapreduce.RowCounter