HBASE-2157 LATEST_TIMESTAMP not replaced by current timestamp in KeyValue

git-svn-id: https://svn.apache.org/repos/asf/hadoop/hbase/trunk@902224 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Michael Stack 2010-01-22 19:06:55 +00:00
parent 60eee421aa
commit 0734674f50
2 changed files with 8 additions and 2 deletions

View File

@ -330,6 +330,8 @@ Release 0.21.0 - Unreleased
HBASE-2151 Remove onelab and include generated thrift classes in javadoc
(Lars Francke via Stack)
HBASE-2149 hbase.regionserver.global.memstore.lowerLimit is too low
HBASE-2157 LATEST_TIMESTAMP not replaced by current timestamp in KeyValue
(bulk loading)
NEW FEATURES
HBASE-1961 HBase EC2 scripts

View File

@ -116,8 +116,12 @@ magnitude faster if not more.
pull from your data source. Your reduce task will need to be hooked up to
{@link org.apache.hadoop.hbase.mapreduce.HFileOutputFormat}. It expects to receive a row id and a value.
The row id must be formatted as a {@link org.apache.hadoop.hbase.io.ImmutableBytesWritable} and the
value as a {@link org.apache.hadoop.hbase.KeyValue} (A KeyValue holds he value for a cell and
its coordinates; row/family/qualifier/timestamp, etc.). Your reduce task
value as a {@link org.apache.hadoop.hbase.KeyValue} (A KeyValue holds the value for a cell and
its coordinates; row/family/qualifier/timestamp, etc.). Note that you must
specify a timestamp when you create the KeyValue in your map task
otherwise the KeyValue will be created with the default LATEST_TIMESTAMP (Long.MAX_VALUE).
Use System.currentTimeMillis() if your data does not inherently bear a timestamp.
Your reduce task
will also need to emit the KeyValues in order. See {@link org.apache.hadoop.hbase.mapreduce.KeyValueSortReducer}
for an example reducer that emits KeyValues in order.
</p>