Java APIs

Newly Added or Modified APIs

• org.apache.hadoop.hbase.Cell of HBase 0.98.3 rather than org.apache.hadoop.hbase.KeyValue of HBase 0.94 is recommended as the key-value data object.
• It is recommended that HConnection connection = HConnectionManager.createConnection(conf) be used to create a connection pool in HBase 0.98.3. The HTablePool is abandoned.
• For details about the new EndPoint API, visit http://hbase.apache.org/book/cp.html.
• The isReversed() and setReversed(boolean reversed) reversed scan methods are added to org.apache.hadoop.hbase.client.Scan.
• For details about API changes from HBase 0.98 to HBase 1.0, visit https://issues.apache.org/jira/browse/hbase-10602.
• org.apache.hadoop.hbase.mapreduce rather than org.apache.hadoop.hbase.mapred is recommended for HBase 1.0.
• For details about the version, visit https://blogs.apache.org/hbase/entry/start_of_a_new_era.
• New APIs added to obtain HBase replication metrics

  Table 1 org.apache.hadoop.hbase.client.replication.ReplicationAdmin

  Method	Description
  getSourceMetricsSummary(String id)	Parameter type: String The source metric summary of the peer ID needs to be obtained. Return type: Map<String, String> Returned: A map, where the key is the RegionServer name and the value is the source cluster metric summary of the specified peer ID. Summary metrics are sizeOfLogToReplicate and timeForLogToReplicate.
  getSourceMetrics(String id)	Parameter type: String The source metric summary of the peer ID needs to be obtained. Return type: Map<String, String> Returned: A map, where the key is the RegionServer name and the value is the source cluster metric of the specified peer ID.
  getSinkMetrics()	Return type: Map<String, String> Returned: A map, where the key is the RegionServer name and the value is the source cluster sink metric of the specified peer ID.
  getPeerSinkMetrics(String id)	Parameter type: String The source metric summary of the peer ID needs to be obtained. Return type: Map<String, String> Returned: A map, where the key is the RegionServer name and the value is the source cluster sink metric of the specified peer ID.

  

  All methods return a Map, where the key is "RegionServer name (IP/Host)" and the value is the string containing all the metrics in format of 'Metric Name'='Metric Value' [, 'Metric Name'= 'Metric Value']*.

  Example: SizeOfHFileRefsQueue=0, AgeOfLastShippedOp=0

  Table 2 org.apache.hadoop.hbase.replication.ReplicationLoadSource

  Method	Description
  getPeerID()	Return type: String Returned: peer cluster ID
  getAgeOfLastShippedOp()	Return type: long Returned: milliseconds that the last successful replication request lasts
  getSizeOfLogQueue()	Return type: long Returned: write-ahead logs (WALs) waiting for replication in the queue
  getTimeStampOfLastShippedOp()	Return type: long Returned: timestamp of the last successful replication request
  getReplicationLag()	Return type: long Returned: interval between current time and the time of the last successful replication request
  getShippedOps()	Return type: long Returned: total number of data ops transferred
  getShippedBytes()	Return type: long Returned: total number of data bytes transferred
  getShippedBatches()	Return type: long Returned: total number of data batches transferred
  getLogReadInBytes()	Return type: long Returned: total number of bytes read from WAL
  getLogEditsRead()	Return type: long Returned: total number of edits read from WAL
  getSizeOfLogToReplicate()	Return type: long Returned: total size of WALs waiting to be replicated in the queue
  getTimeForLogToReplicate()	Return type: long Returned: seconds spent in replicating WALs in the queue
  getShippedHFiles()	Return type: long Returned: total number of HFiles transferred
  getSizeOfHFileRefsQueue()	Return type: long Returned: total number of HFiles waiting to be replicated
  getLogEditsFiltered()	Return type: long Returned: total number of WAL edits filtered
  getFailedReplicationAttempts()	Return type: long Returned: times failed to replicate data for a single request

  Table 3 org.apache.hadoop.hbase.replication.ReplicationLoadSink

  Method	Description
  getAgeOfLastAppliedOp()	Return type: long Returned: milliseconds that the last successful applied WAL edits last
  getTimeStampsOfLastAppliedOp()	Return type: long Returned: timestamp of the last successful applied WAL edit
  getAppliedBatches()	Return type: long Returned: total number of data batches applied
  getAppliedOps()	Return type: long Returned: total number of data ops applied
  getAppliedHFiles()	Return type: long Returned: total number of HFiles applied

  

  The new API OF Replication Admin obtains the metric values from HMaster. Each RegionServer reports status to HMaster at every heartbeat interval, which is 3 seconds by default. Therefore, this API reports the latest metric value at the last heartbeat by using the RegionServer.

  If you need the latest metric value, use the JMX API provided by the RegionServer.

• 1.3.1 (MRS 1.9.2) API Changes
  • Added HIndex API

  Table 4 org.apache.hadoop.hbase.hindex.client.HIndexAdmin

  Method	Description
  addIndices(TableName tablename,TableIndices tableIndices)	Parameter: TableName Name of the table to which the user wants to add a specified index. Parameter: TableIndices Table index to be added to the table Return type: void
  addIndicesWithData(TableName tablename,TableIndices tableIndices)	Parameter: TableName Name of the table to which the user wants to add a specified index Parameter: TableIndices Table index to be added to the table Return type: void
  dropIndices(TableName tableName,List <String> list)	Parameter: TableName Name of the table from which the user wants to delete an index Parameter: List<String> Contains the list of indexes to be deleted. Return type: void
  dropIndicesWithData(TableName tableName,List <String> list)	Parameter: TableName Name of the table from which the user wants to delete a specified index Parameter: List<String> Contains the list of indexes to be deleted. Return type: void
  disableIndices(TableName tableName,List <String> list)	Parameter: TableName Name of the table for which the user wants to disable a specified index Parameter: List<String> Contains the list of indexes to be disabled. Return type: void
  enableIndices(TableName tableName,List <String> list)	Parameter: TableName Name of the table for which the user wants to enable a specified index Parameter: List<String> Contains the list of indexes to be enabled. Return type: void
  listIndices(TableName tableName)	Parameter: TableName Name of the table for which the user wants to list all indexes Return type: List <Pair <HIndexSpecification,IndexState >> Return: A secondary index list is returned. The first element is the index specification, and the second element is the current state of the index.