http://git-wip-us.apache.org/repos/asf/hbase-site/blob/c7e84622/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html index ef7c1e1..a541dfa 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html @@ -26,51 +26,51 @@ <span class="sourceLineNo">018</span> */<a name="line.18"></a> <span class="sourceLineNo">019</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.19"></a> <span class="sourceLineNo">020</span><a name="line.20"></a> -<span class="sourceLineNo">021</span>import com.google.protobuf.InvalidProtocolBufferException;<a name="line.21"></a> -<span class="sourceLineNo">022</span>import com.codahale.metrics.MetricRegistry;<a name="line.22"></a> -<span class="sourceLineNo">023</span>import org.apache.commons.logging.Log;<a name="line.23"></a> -<span class="sourceLineNo">024</span>import org.apache.commons.logging.LogFactory;<a name="line.24"></a> -<span class="sourceLineNo">025</span>import org.apache.hadoop.conf.Configuration;<a name="line.25"></a> -<span class="sourceLineNo">026</span>import org.apache.hadoop.fs.FileSystem;<a name="line.26"></a> -<span class="sourceLineNo">027</span>import org.apache.hadoop.fs.Path;<a name="line.27"></a> -<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.28"></a> -<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.HConstants;<a name="line.29"></a> -<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.MetaTableAccessor;<a name="line.30"></a> -<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.TableName;<a name="line.31"></a> -<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.32"></a> -<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.classification.InterfaceStability;<a name="line.33"></a> -<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.34"></a> -<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.35"></a> -<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.client.Put;<a name="line.36"></a> -<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.37"></a> -<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.38"></a> -<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.protobuf.ProtobufUtil;<a name="line.39"></a> -<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.protobuf.generated.ClientProtos;<a name="line.40"></a> -<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.security.User;<a name="line.41"></a> -<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.security.UserProvider;<a name="line.42"></a> -<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.security.token.TokenUtil;<a name="line.43"></a> -<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.util.Base64;<a name="line.44"></a> -<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.45"></a> -<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.zookeeper.ZKConfig;<a name="line.46"></a> -<span class="sourceLineNo">047</span>import org.apache.hadoop.io.Writable;<a name="line.47"></a> -<span class="sourceLineNo">048</span>import org.apache.hadoop.mapreduce.InputFormat;<a name="line.48"></a> -<span class="sourceLineNo">049</span>import org.apache.hadoop.mapreduce.Job;<a name="line.49"></a> -<span class="sourceLineNo">050</span>import org.apache.hadoop.util.StringUtils;<a name="line.50"></a> -<span class="sourceLineNo">051</span><a name="line.51"></a> -<span class="sourceLineNo">052</span>import java.io.File;<a name="line.52"></a> -<span class="sourceLineNo">053</span>import java.io.IOException;<a name="line.53"></a> -<span class="sourceLineNo">054</span>import java.net.URL;<a name="line.54"></a> -<span class="sourceLineNo">055</span>import java.net.URLDecoder;<a name="line.55"></a> -<span class="sourceLineNo">056</span>import java.util.ArrayList;<a name="line.56"></a> -<span class="sourceLineNo">057</span>import java.util.Collection;<a name="line.57"></a> -<span class="sourceLineNo">058</span>import java.util.Enumeration;<a name="line.58"></a> -<span class="sourceLineNo">059</span>import java.util.HashMap;<a name="line.59"></a> -<span class="sourceLineNo">060</span>import java.util.HashSet;<a name="line.60"></a> -<span class="sourceLineNo">061</span>import java.util.List;<a name="line.61"></a> -<span class="sourceLineNo">062</span>import java.util.Map;<a name="line.62"></a> -<span class="sourceLineNo">063</span>import java.util.Set;<a name="line.63"></a> -<span class="sourceLineNo">064</span>import java.util.zip.ZipEntry;<a name="line.64"></a> -<span class="sourceLineNo">065</span>import java.util.zip.ZipFile;<a name="line.65"></a> +<span class="sourceLineNo">021</span>import java.io.File;<a name="line.21"></a> +<span class="sourceLineNo">022</span>import java.io.IOException;<a name="line.22"></a> +<span class="sourceLineNo">023</span>import java.net.URL;<a name="line.23"></a> +<span class="sourceLineNo">024</span>import java.net.URLDecoder;<a name="line.24"></a> +<span class="sourceLineNo">025</span>import java.util.ArrayList;<a name="line.25"></a> +<span class="sourceLineNo">026</span>import java.util.Collection;<a name="line.26"></a> +<span class="sourceLineNo">027</span>import java.util.Enumeration;<a name="line.27"></a> +<span class="sourceLineNo">028</span>import java.util.HashMap;<a name="line.28"></a> +<span class="sourceLineNo">029</span>import java.util.HashSet;<a name="line.29"></a> +<span class="sourceLineNo">030</span>import java.util.List;<a name="line.30"></a> +<span class="sourceLineNo">031</span>import java.util.Map;<a name="line.31"></a> +<span class="sourceLineNo">032</span>import java.util.Set;<a name="line.32"></a> +<span class="sourceLineNo">033</span>import java.util.zip.ZipEntry;<a name="line.33"></a> +<span class="sourceLineNo">034</span>import java.util.zip.ZipFile;<a name="line.34"></a> +<span class="sourceLineNo">035</span><a name="line.35"></a> +<span class="sourceLineNo">036</span>import org.apache.commons.logging.Log;<a name="line.36"></a> +<span class="sourceLineNo">037</span>import org.apache.commons.logging.LogFactory;<a name="line.37"></a> +<span class="sourceLineNo">038</span>import org.apache.hadoop.conf.Configuration;<a name="line.38"></a> +<span class="sourceLineNo">039</span>import org.apache.hadoop.fs.FileSystem;<a name="line.39"></a> +<span class="sourceLineNo">040</span>import org.apache.hadoop.fs.Path;<a name="line.40"></a> +<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.41"></a> +<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.HConstants;<a name="line.42"></a> +<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.MetaTableAccessor;<a name="line.43"></a> +<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.TableName;<a name="line.44"></a> +<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.45"></a> +<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.classification.InterfaceStability;<a name="line.46"></a> +<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.47"></a> +<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.48"></a> +<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.Put;<a name="line.49"></a> +<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.50"></a> +<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.51"></a> +<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil;<a name="line.52"></a> +<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.ClientProtos;<a name="line.53"></a> +<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.security.User;<a name="line.54"></a> +<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.security.UserProvider;<a name="line.55"></a> +<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.security.token.TokenUtil;<a name="line.56"></a> +<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.util.Base64;<a name="line.57"></a> +<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.58"></a> +<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.zookeeper.ZKConfig;<a name="line.59"></a> +<span class="sourceLineNo">060</span>import org.apache.hadoop.io.Writable;<a name="line.60"></a> +<span class="sourceLineNo">061</span>import org.apache.hadoop.mapreduce.InputFormat;<a name="line.61"></a> +<span class="sourceLineNo">062</span>import org.apache.hadoop.mapreduce.Job;<a name="line.62"></a> +<span class="sourceLineNo">063</span>import org.apache.hadoop.util.StringUtils;<a name="line.63"></a> +<span class="sourceLineNo">064</span><a name="line.64"></a> +<span class="sourceLineNo">065</span>import com.codahale.metrics.MetricRegistry;<a name="line.65"></a> <span class="sourceLineNo">066</span><a name="line.66"></a> <span class="sourceLineNo">067</span>/**<a name="line.67"></a> <span class="sourceLineNo">068</span> * Utility for {@link TableMapper} and {@link TableReducer}<a name="line.68"></a> @@ -583,461 +583,454 @@ <span class="sourceLineNo">575</span> */<a name="line.575"></a> <span class="sourceLineNo">576</span> public static Scan convertStringToScan(String base64) throws IOException {<a name="line.576"></a> <span class="sourceLineNo">577</span> byte [] decoded = Base64.decode(base64);<a name="line.577"></a> -<span class="sourceLineNo">578</span> ClientProtos.Scan scan;<a name="line.578"></a> -<span class="sourceLineNo">579</span> try {<a name="line.579"></a> -<span class="sourceLineNo">580</span> scan = ClientProtos.Scan.parseFrom(decoded);<a name="line.580"></a> -<span class="sourceLineNo">581</span> } catch (InvalidProtocolBufferException ipbe) {<a name="line.581"></a> -<span class="sourceLineNo">582</span> throw new IOException(ipbe);<a name="line.582"></a> -<span class="sourceLineNo">583</span> }<a name="line.583"></a> -<span class="sourceLineNo">584</span><a name="line.584"></a> -<span class="sourceLineNo">585</span> return ProtobufUtil.toScan(scan);<a name="line.585"></a> -<span class="sourceLineNo">586</span> }<a name="line.586"></a> -<span class="sourceLineNo">587</span><a name="line.587"></a> -<span class="sourceLineNo">588</span> /**<a name="line.588"></a> -<span class="sourceLineNo">589</span> * Use this before submitting a TableReduce job. It will<a name="line.589"></a> -<span class="sourceLineNo">590</span> * appropriately set up the JobConf.<a name="line.590"></a> -<span class="sourceLineNo">591</span> *<a name="line.591"></a> -<span class="sourceLineNo">592</span> * @param table The output table.<a name="line.592"></a> -<span class="sourceLineNo">593</span> * @param reducer The reducer class to use.<a name="line.593"></a> -<span class="sourceLineNo">594</span> * @param job The current job to adjust.<a name="line.594"></a> -<span class="sourceLineNo">595</span> * @throws IOException When determining the region count fails.<a name="line.595"></a> -<span class="sourceLineNo">596</span> */<a name="line.596"></a> -<span class="sourceLineNo">597</span> public static void initTableReducerJob(String table,<a name="line.597"></a> -<span class="sourceLineNo">598</span> Class<? extends TableReducer> reducer, Job job)<a name="line.598"></a> -<span class="sourceLineNo">599</span> throws IOException {<a name="line.599"></a> -<span class="sourceLineNo">600</span> initTableReducerJob(table, reducer, job, null);<a name="line.600"></a> -<span class="sourceLineNo">601</span> }<a name="line.601"></a> -<span class="sourceLineNo">602</span><a name="line.602"></a> -<span class="sourceLineNo">603</span> /**<a name="line.603"></a> -<span class="sourceLineNo">604</span> * Use this before submitting a TableReduce job. It will<a name="line.604"></a> -<span class="sourceLineNo">605</span> * appropriately set up the JobConf.<a name="line.605"></a> -<span class="sourceLineNo">606</span> *<a name="line.606"></a> -<span class="sourceLineNo">607</span> * @param table The output table.<a name="line.607"></a> -<span class="sourceLineNo">608</span> * @param reducer The reducer class to use.<a name="line.608"></a> -<span class="sourceLineNo">609</span> * @param job The current job to adjust.<a name="line.609"></a> -<span class="sourceLineNo">610</span> * @param partitioner Partitioner to use. Pass <code>null</code> to use<a name="line.610"></a> -<span class="sourceLineNo">611</span> * default partitioner.<a name="line.611"></a> -<span class="sourceLineNo">612</span> * @throws IOException When determining the region count fails.<a name="line.612"></a> -<span class="sourceLineNo">613</span> */<a name="line.613"></a> -<span class="sourceLineNo">614</span> public static void initTableReducerJob(String table,<a name="line.614"></a> -<span class="sourceLineNo">615</span> Class<? extends TableReducer> reducer, Job job,<a name="line.615"></a> -<span class="sourceLineNo">616</span> Class partitioner) throws IOException {<a name="line.616"></a> -<span class="sourceLineNo">617</span> initTableReducerJob(table, reducer, job, partitioner, null, null, null);<a name="line.617"></a> -<span class="sourceLineNo">618</span> }<a name="line.618"></a> -<span class="sourceLineNo">619</span><a name="line.619"></a> -<span class="sourceLineNo">620</span> /**<a name="line.620"></a> -<span class="sourceLineNo">621</span> * Use this before submitting a TableReduce job. It will<a name="line.621"></a> -<span class="sourceLineNo">622</span> * appropriately set up the JobConf.<a name="line.622"></a> -<span class="sourceLineNo">623</span> *<a name="line.623"></a> -<span class="sourceLineNo">624</span> * @param table The output table.<a name="line.624"></a> -<span class="sourceLineNo">625</span> * @param reducer The reducer class to use.<a name="line.625"></a> -<span class="sourceLineNo">626</span> * @param job The current job to adjust. Make sure the passed job is<a name="line.626"></a> -<span class="sourceLineNo">627</span> * carrying all necessary HBase configuration.<a name="line.627"></a> -<span class="sourceLineNo">628</span> * @param partitioner Partitioner to use. Pass <code>null</code> to use<a name="line.628"></a> -<span class="sourceLineNo">629</span> * default partitioner.<a name="line.629"></a> -<span class="sourceLineNo">630</span> * @param quorumAddress Distant cluster to write to; default is null for<a name="line.630"></a> -<span class="sourceLineNo">631</span> * output to the cluster that is designated in <code>hbase-site.xml</code>.<a name="line.631"></a> -<span class="sourceLineNo">632</span> * Set this String to the zookeeper ensemble of an alternate remote cluster<a name="line.632"></a> -<span class="sourceLineNo">633</span> * when you would have the reduce write a cluster that is other than the<a name="line.633"></a> -<span class="sourceLineNo">634</span> * default; e.g. copying tables between clusters, the source would be<a name="line.634"></a> -<span class="sourceLineNo">635</span> * designated by <code>hbase-site.xml</code> and this param would have the<a name="line.635"></a> -<span class="sourceLineNo">636</span> * ensemble address of the remote cluster. The format to pass is particular.<a name="line.636"></a> -<span class="sourceLineNo">637</span> * Pass <code> &lt;hbase.zookeeper.quorum&gt;:&lt;<a name="line.637"></a> -<span class="sourceLineNo">638</span> * hbase.zookeeper.client.port&gt;:&lt;zookeeper.znode.parent&gt;<a name="line.638"></a> -<span class="sourceLineNo">639</span> * </code> such as <code>server,server2,server3:2181:/hbase</code>.<a name="line.639"></a> -<span class="sourceLineNo">640</span> * @param serverClass redefined hbase.regionserver.class<a name="line.640"></a> -<span class="sourceLineNo">641</span> * @param serverImpl redefined hbase.regionserver.impl<a name="line.641"></a> -<span class="sourceLineNo">642</span> * @throws IOException When determining the region count fails.<a name="line.642"></a> -<span class="sourceLineNo">643</span> */<a name="line.643"></a> -<span class="sourceLineNo">644</span> public static void initTableReducerJob(String table,<a name="line.644"></a> -<span class="sourceLineNo">645</span> Class<? extends TableReducer> reducer, Job job,<a name="line.645"></a> -<span class="sourceLineNo">646</span> Class partitioner, String quorumAddress, String serverClass,<a name="line.646"></a> -<span class="sourceLineNo">647</span> String serverImpl) throws IOException {<a name="line.647"></a> -<span class="sourceLineNo">648</span> initTableReducerJob(table, reducer, job, partitioner, quorumAddress,<a name="line.648"></a> -<span class="sourceLineNo">649</span> serverClass, serverImpl, true);<a name="line.649"></a> -<span class="sourceLineNo">650</span> }<a name="line.650"></a> -<span class="sourceLineNo">651</span><a name="line.651"></a> -<span class="sourceLineNo">652</span> /**<a name="line.652"></a> -<span class="sourceLineNo">653</span> * Use this before submitting a TableReduce job. It will<a name="line.653"></a> -<span class="sourceLineNo">654</span> * appropriately set up the JobConf.<a name="line.654"></a> -<span class="sourceLineNo">655</span> *<a name="line.655"></a> -<span class="sourceLineNo">656</span> * @param table The output table.<a name="line.656"></a> -<span class="sourceLineNo">657</span> * @param reducer The reducer class to use.<a name="line.657"></a> -<span class="sourceLineNo">658</span> * @param job The current job to adjust. Make sure the passed job is<a name="line.658"></a> -<span class="sourceLineNo">659</span> * carrying all necessary HBase configuration.<a name="line.659"></a> -<span class="sourceLineNo">660</span> * @param partitioner Partitioner to use. Pass <code>null</code> to use<a name="line.660"></a> -<span class="sourceLineNo">661</span> * default partitioner.<a name="line.661"></a> -<span class="sourceLineNo">662</span> * @param quorumAddress Distant cluster to write to; default is null for<a name="line.662"></a> -<span class="sourceLineNo">663</span> * output to the cluster that is designated in <code>hbase-site.xml</code>.<a name="line.663"></a> -<span class="sourceLineNo">664</span> * Set this String to the zookeeper ensemble of an alternate remote cluster<a name="line.664"></a> -<span class="sourceLineNo">665</span> * when you would have the reduce write a cluster that is other than the<a name="line.665"></a> -<span class="sourceLineNo">666</span> * default; e.g. copying tables between clusters, the source would be<a name="line.666"></a> -<span class="sourceLineNo">667</span> * designated by <code>hbase-site.xml</code> and this param would have the<a name="line.667"></a> -<span class="sourceLineNo">668</span> * ensemble address of the remote cluster. The format to pass is particular.<a name="line.668"></a> -<span class="sourceLineNo">669</span> * Pass <code> &lt;hbase.zookeeper.quorum&gt;:&lt;<a name="line.669"></a> -<span class="sourceLineNo">670</span> * hbase.zookeeper.client.port&gt;:&lt;zookeeper.znode.parent&gt;<a name="line.670"></a> -<span class="sourceLineNo">671</span> * </code> such as <code>server,server2,server3:2181:/hbase</code>.<a name="line.671"></a> -<span class="sourceLineNo">672</span> * @param serverClass redefined hbase.regionserver.class<a name="line.672"></a> -<span class="sourceLineNo">673</span> * @param serverImpl redefined hbase.regionserver.impl<a name="line.673"></a> -<span class="sourceLineNo">674</span> * @param addDependencyJars upload HBase jars and jars for any of the configured<a name="line.674"></a> -<span class="sourceLineNo">675</span> * job classes via the distributed cache (tmpjars).<a name="line.675"></a> -<span class="sourceLineNo">676</span> * @throws IOException When determining the region count fails.<a name="line.676"></a> -<span class="sourceLineNo">677</span> */<a name="line.677"></a> -<span class="sourceLineNo">678</span> public static void initTableReducerJob(String table,<a name="line.678"></a> -<span class="sourceLineNo">679</span> Class<? extends TableReducer> reducer, Job job,<a name="line.679"></a> -<span class="sourceLineNo">680</span> Class partitioner, String quorumAddress, String serverClass,<a name="line.680"></a> -<span class="sourceLineNo">681</span> String serverImpl, boolean addDependencyJars) throws IOException {<a name="line.681"></a> -<span class="sourceLineNo">682</span><a name="line.682"></a> -<span class="sourceLineNo">683</span> Configuration conf = job.getConfiguration();<a name="line.683"></a> -<span class="sourceLineNo">684</span> HBaseConfiguration.merge(conf, HBaseConfiguration.create(conf));<a name="line.684"></a> -<span class="sourceLineNo">685</span> job.setOutputFormatClass(TableOutputFormat.class);<a name="line.685"></a> -<span class="sourceLineNo">686</span> if (reducer != null) job.setReducerClass(reducer);<a name="line.686"></a> -<span class="sourceLineNo">687</span> conf.set(TableOutputFormat.OUTPUT_TABLE, table);<a name="line.687"></a> -<span class="sourceLineNo">688</span> conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.688"></a> -<span class="sourceLineNo">689</span> MutationSerialization.class.getName(), ResultSerialization.class.getName());<a name="line.689"></a> -<span class="sourceLineNo">690</span> // If passed a quorum/ensemble address, pass it on to TableOutputFormat.<a name="line.690"></a> -<span class="sourceLineNo">691</span> if (quorumAddress != null) {<a name="line.691"></a> -<span class="sourceLineNo">692</span> // Calling this will validate the format<a name="line.692"></a> -<span class="sourceLineNo">693</span> ZKConfig.validateClusterKey(quorumAddress);<a name="line.693"></a> -<span class="sourceLineNo">694</span> conf.set(TableOutputFormat.QUORUM_ADDRESS,quorumAddress);<a name="line.694"></a> -<span class="sourceLineNo">695</span> }<a name="line.695"></a> -<span class="sourceLineNo">696</span> if (serverClass != null && serverImpl != null) {<a name="line.696"></a> -<span class="sourceLineNo">697</span> conf.set(TableOutputFormat.REGION_SERVER_CLASS, serverClass);<a name="line.697"></a> -<span class="sourceLineNo">698</span> conf.set(TableOutputFormat.REGION_SERVER_IMPL, serverImpl);<a name="line.698"></a> -<span class="sourceLineNo">699</span> }<a name="line.699"></a> -<span class="sourceLineNo">700</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.700"></a> -<span class="sourceLineNo">701</span> job.setOutputValueClass(Writable.class);<a name="line.701"></a> -<span class="sourceLineNo">702</span> if (partitioner == HRegionPartitioner.class) {<a name="line.702"></a> -<span class="sourceLineNo">703</span> job.setPartitionerClass(HRegionPartitioner.class);<a name="line.703"></a> -<span class="sourceLineNo">704</span> int regions = MetaTableAccessor.getRegionCount(conf, TableName.valueOf(table));<a name="line.704"></a> -<span class="sourceLineNo">705</span> if (job.getNumReduceTasks() > regions) {<a name="line.705"></a> -<span class="sourceLineNo">706</span> job.setNumReduceTasks(regions);<a name="line.706"></a> -<span class="sourceLineNo">707</span> }<a name="line.707"></a> -<span class="sourceLineNo">708</span> } else if (partitioner != null) {<a name="line.708"></a> -<span class="sourceLineNo">709</span> job.setPartitionerClass(partitioner);<a name="line.709"></a> -<span class="sourceLineNo">710</span> }<a name="line.710"></a> +<span class="sourceLineNo">578</span> return ProtobufUtil.toScan(ClientProtos.Scan.parseFrom(decoded));<a name="line.578"></a> +<span class="sourceLineNo">579</span> }<a name="line.579"></a> +<span class="sourceLineNo">580</span><a name="line.580"></a> +<span class="sourceLineNo">581</span> /**<a name="line.581"></a> +<span class="sourceLineNo">582</span> * Use this before submitting a TableReduce job. It will<a name="line.582"></a> +<span class="sourceLineNo">583</span> * appropriately set up the JobConf.<a name="line.583"></a> +<span class="sourceLineNo">584</span> *<a name="line.584"></a> +<span class="sourceLineNo">585</span> * @param table The output table.<a name="line.585"></a> +<span class="sourceLineNo">586</span> * @param reducer The reducer class to use.<a name="line.586"></a> +<span class="sourceLineNo">587</span> * @param job The current job to adjust.<a name="line.587"></a> +<span class="sourceLineNo">588</span> * @throws IOException When determining the region count fails.<a name="line.588"></a> +<span class="sourceLineNo">589</span> */<a name="line.589"></a> +<span class="sourceLineNo">590</span> public static void initTableReducerJob(String table,<a name="line.590"></a> +<span class="sourceLineNo">591</span> Class<? extends TableReducer> reducer, Job job)<a name="line.591"></a> +<span class="sourceLineNo">592</span> throws IOException {<a name="line.592"></a> +<span class="sourceLineNo">593</span> initTableReducerJob(table, reducer, job, null);<a name="line.593"></a> +<span class="sourceLineNo">594</span> }<a name="line.594"></a> +<span class="sourceLineNo">595</span><a name="line.595"></a> +<span class="sourceLineNo">596</span> /**<a name="line.596"></a> +<span class="sourceLineNo">597</span> * Use this before submitting a TableReduce job. It will<a name="line.597"></a> +<span class="sourceLineNo">598</span> * appropriately set up the JobConf.<a name="line.598"></a> +<span class="sourceLineNo">599</span> *<a name="line.599"></a> +<span class="sourceLineNo">600</span> * @param table The output table.<a name="line.600"></a> +<span class="sourceLineNo">601</span> * @param reducer The reducer class to use.<a name="line.601"></a> +<span class="sourceLineNo">602</span> * @param job The current job to adjust.<a name="line.602"></a> +<span class="sourceLineNo">603</span> * @param partitioner Partitioner to use. Pass <code>null</code> to use<a name="line.603"></a> +<span class="sourceLineNo">604</span> * default partitioner.<a name="line.604"></a> +<span class="sourceLineNo">605</span> * @throws IOException When determining the region count fails.<a name="line.605"></a> +<span class="sourceLineNo">606</span> */<a name="line.606"></a> +<span class="sourceLineNo">607</span> public static void initTableReducerJob(String table,<a name="line.607"></a> +<span class="sourceLineNo">608</span> Class<? extends TableReducer> reducer, Job job,<a name="line.608"></a> +<span class="sourceLineNo">609</span> Class partitioner) throws IOException {<a name="line.609"></a> +<span class="sourceLineNo">610</span> initTableReducerJob(table, reducer, job, partitioner, null, null, null);<a name="line.610"></a> +<span class="sourceLineNo">611</span> }<a name="line.611"></a> +<span class="sourceLineNo">612</span><a name="line.612"></a> +<span class="sourceLineNo">613</span> /**<a name="line.613"></a> +<span class="sourceLineNo">614</span> * Use this before submitting a TableReduce job. It will<a name="line.614"></a> +<span class="sourceLineNo">615</span> * appropriately set up the JobConf.<a name="line.615"></a> +<span class="sourceLineNo">616</span> *<a name="line.616"></a> +<span class="sourceLineNo">617</span> * @param table The output table.<a name="line.617"></a> +<span class="sourceLineNo">618</span> * @param reducer The reducer class to use.<a name="line.618"></a> +<span class="sourceLineNo">619</span> * @param job The current job to adjust. Make sure the passed job is<a name="line.619"></a> +<span class="sourceLineNo">620</span> * carrying all necessary HBase configuration.<a name="line.620"></a> +<span class="sourceLineNo">621</span> * @param partitioner Partitioner to use. Pass <code>null</code> to use<a name="line.621"></a> +<span class="sourceLineNo">622</span> * default partitioner.<a name="line.622"></a> +<span class="sourceLineNo">623</span> * @param quorumAddress Distant cluster to write to; default is null for<a name="line.623"></a> +<span class="sourceLineNo">624</span> * output to the cluster that is designated in <code>hbase-site.xml</code>.<a name="line.624"></a> +<span class="sourceLineNo">625</span> * Set this String to the zookeeper ensemble of an alternate remote cluster<a name="line.625"></a> +<span class="sourceLineNo">626</span> * when you would have the reduce write a cluster that is other than the<a name="line.626"></a> +<span class="sourceLineNo">627</span> * default; e.g. copying tables between clusters, the source would be<a name="line.627"></a> +<span class="sourceLineNo">628</span> * designated by <code>hbase-site.xml</code> and this param would have the<a name="line.628"></a> +<span class="sourceLineNo">629</span> * ensemble address of the remote cluster. The format to pass is particular.<a name="line.629"></a> +<span class="sourceLineNo">630</span> * Pass <code> &lt;hbase.zookeeper.quorum&gt;:&lt;<a name="line.630"></a> +<span class="sourceLineNo">631</span> * hbase.zookeeper.client.port&gt;:&lt;zookeeper.znode.parent&gt;<a name="line.631"></a> +<span class="sourceLineNo">632</span> * </code> such as <code>server,server2,server3:2181:/hbase</code>.<a name="line.632"></a> +<span class="sourceLineNo">633</span> * @param serverClass redefined hbase.regionserver.class<a name="line.633"></a> +<span class="sourceLineNo">634</span> * @param serverImpl redefined hbase.regionserver.impl<a name="line.634"></a> +<span class="sourceLineNo">635</span> * @throws IOException When determining the region count fails.<a name="line.635"></a> +<span class="sourceLineNo">636</span> */<a name="line.636"></a> +<span class="sourceLineNo">637</span> public static void initTableReducerJob(String table,<a name="line.637"></a> +<span class="sourceLineNo">638</span> Class<? extends TableReducer> reducer, Job job,<a name="line.638"></a> +<span class="sourceLineNo">639</span> Class partitioner, String quorumAddress, String serverClass,<a name="line.639"></a> +<span class="sourceLineNo">640</span> String serverImpl) throws IOException {<a name="line.640"></a> +<span class="sourceLineNo">641</span> initTableReducerJob(table, reducer, job, partitioner, quorumAddress,<a name="line.641"></a> +<span class="sourceLineNo">642</span> serverClass, serverImpl, true);<a name="line.642"></a> +<span class="sourceLineNo">643</span> }<a name="line.643"></a> +<span class="sourceLineNo">644</span><a name="line.644"></a> +<span class="sourceLineNo">645</span> /**<a name="line.645"></a> +<span class="sourceLineNo">646</span> * Use this before submitting a TableReduce job. It will<a name="line.646"></a> +<span class="sourceLineNo">647</span> * appropriately set up the JobConf.<a name="line.647"></a> +<span class="sourceLineNo">648</span> *<a name="line.648"></a> +<span class="sourceLineNo">649</span> * @param table The output table.<a name="line.649"></a> +<span class="sourceLineNo">650</span> * @param reducer The reducer class to use.<a name="line.650"></a> +<span class="sourceLineNo">651</span> * @param job The current job to adjust. Make sure the passed job is<a name="line.651"></a> +<span class="sourceLineNo">652</span> * carrying all necessary HBase configuration.<a name="line.652"></a> +<span class="sourceLineNo">653</span> * @param partitioner Partitioner to use. Pass <code>null</code> to use<a name="line.653"></a> +<span class="sourceLineNo">654</span> * default partitioner.<a name="line.654"></a> +<span class="sourceLineNo">655</span> * @param quorumAddress Distant cluster to write to; default is null for<a name="line.655"></a> +<span class="sourceLineNo">656</span> * output to the cluster that is designated in <code>hbase-site.xml</code>.<a name="line.656"></a> +<span class="sourceLineNo">657</span> * Set this String to the zookeeper ensemble of an alternate remote cluster<a name="line.657"></a> +<span class="sourceLineNo">658</span> * when you would have the reduce write a cluster that is other than the<a name="line.658"></a> +<span class="sourceLineNo">659</span> * default; e.g. copying tables between clusters, the source would be<a name="line.659"></a> +<span class="sourceLineNo">660</span> * designated by <code>hbase-site.xml</code> and this param would have the<a name="line.660"></a> +<span class="sourceLineNo">661</span> * ensemble address of the remote cluster. The format to pass is particular.<a name="line.661"></a> +<span class="sourceLineNo">662</span> * Pass <code> &lt;hbase.zookeeper.quorum&gt;:&lt;<a name="line.662"></a> +<span class="sourceLineNo">663</span> * hbase.zookeeper.client.port&gt;:&lt;zookeeper.znode.parent&gt;<a name="line.663"></a> +<span class="sourceLineNo">664</span> * </code> such as <code>server,server2,server3:2181:/hbase</code>.<a name="line.664"></a> +<span class="sourceLineNo">665</span> * @param serverClass redefined hbase.regionserver.class<a name="line.665"></a> +<span class="sourceLineNo">666</span> * @param serverImpl redefined hbase.regionserver.impl<a name="line.666"></a> +<span class="sourceLineNo">667</span> * @param addDependencyJars upload HBase jars and jars for any of the configured<a name="line.667"></a> +<span class="sourceLineNo">668</span> * job classes via the distributed cache (tmpjars).<a name="line.668"></a> +<span class="sourceLineNo">669</span> * @throws IOException When determining the region count fails.<a name="line.669"></a> +<span class="sourceLineNo">670</span> */<a name="line.670"></a> +<span class="sourceLineNo">671</span> public static void initTableReducerJob(String table,<a name="line.671"></a> +<span class="sourceLineNo">672</span> Class<? extends TableReducer> reducer, Job job,<a name="line.672"></a> +<span class="sourceLineNo">673</span> Class partitioner, String quorumAddress, String serverClass,<a name="line.673"></a> +<span class="sourceLineNo">674</span> String serverImpl, boolean addDependencyJars) throws IOException {<a name="line.674"></a> +<span class="sourceLineNo">675</span><a name="line.675"></a> +<span class="sourceLineNo">676</span> Configuration conf = job.getConfiguration();<a name="line.676"></a> +<span class="sourceLineNo">677</span> HBaseConfiguration.merge(conf, HBaseConfiguration.create(conf));<a name="line.677"></a> +<span class="sourceLineNo">678</span> job.setOutputFormatClass(TableOutputFormat.class);<a name="line.678"></a> +<span class="sourceLineNo">679</span> if (reducer != null) job.setReducerClass(reducer);<a name="line.679"></a> +<span class="sourceLineNo">680</span> conf.set(TableOutputFormat.OUTPUT_TABLE, table);<a name="line.680"></a> +<span class="sourceLineNo">681</span> conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.681"></a> +<span class="sourceLineNo">682</span> MutationSerialization.class.getName(), ResultSerialization.class.getName());<a name="line.682"></a> +<span class="sourceLineNo">683</span> // If passed a quorum/ensemble address, pass it on to TableOutputFormat.<a name="line.683"></a> +<span class="sourceLineNo">684</span> if (quorumAddress != null) {<a name="line.684"></a> +<span class="sourceLineNo">685</span> // Calling this will validate the format<a name="line.685"></a> +<span class="sourceLineNo">686</span> ZKConfig.validateClusterKey(quorumAddress);<a name="line.686"></a> +<span class="sourceLineNo">687</span> conf.set(TableOutputFormat.QUORUM_ADDRESS,quorumAddress);<a name="line.687"></a> +<span class="sourceLineNo">688</span> }<a name="line.688"></a> +<span class="sourceLineNo">689</span> if (serverClass != null && serverImpl != null) {<a name="line.689"></a> +<span class="sourceLineNo">690</span> conf.set(TableOutputFormat.REGION_SERVER_CLASS, serverClass);<a name="line.690"></a> +<span class="sourceLineNo">691</span> conf.set(TableOutputFormat.REGION_SERVER_IMPL, serverImpl);<a name="line.691"></a> +<span class="sourceLineNo">692</span> }<a name="line.692"></a> +<span class="sourceLineNo">693</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.693"></a> +<span class="sourceLineNo">694</span> job.setOutputValueClass(Writable.class);<a name="line.694"></a> +<span class="sourceLineNo">695</span> if (partitioner == HRegionPartitioner.class) {<a name="line.695"></a> +<span class="sourceLineNo">696</span> job.setPartitionerClass(HRegionPartitioner.class);<a name="line.696"></a> +<span class="sourceLineNo">697</span> int regions = MetaTableAccessor.getRegionCount(conf, TableName.valueOf(table));<a name="line.697"></a> +<span class="sourceLineNo">698</span> if (job.getNumReduceTasks() > regions) {<a name="line.698"></a> +<span class="sourceLineNo">699</span> job.setNumReduceTasks(regions);<a name="line.699"></a> +<span class="sourceLineNo">700</span> }<a name="line.700"></a> +<span class="sourceLineNo">701</span> } else if (partitioner != null) {<a name="line.701"></a> +<span class="sourceLineNo">702</span> job.setPartitionerClass(partitioner);<a name="line.702"></a> +<span class="sourceLineNo">703</span> }<a name="line.703"></a> +<span class="sourceLineNo">704</span><a name="line.704"></a> +<span class="sourceLineNo">705</span> if (addDependencyJars) {<a name="line.705"></a> +<span class="sourceLineNo">706</span> addDependencyJars(job);<a name="line.706"></a> +<span class="sourceLineNo">707</span> }<a name="line.707"></a> +<span class="sourceLineNo">708</span><a name="line.708"></a> +<span class="sourceLineNo">709</span> initCredentials(job);<a name="line.709"></a> +<span class="sourceLineNo">710</span> }<a name="line.710"></a> <span class="sourceLineNo">711</span><a name="line.711"></a> -<span class="sourceLineNo">712</span> if (addDependencyJars) {<a name="line.712"></a> -<span class="sourceLineNo">713</span> addDependencyJars(job);<a name="line.713"></a> -<span class="sourceLineNo">714</span> }<a name="line.714"></a> -<span class="sourceLineNo">715</span><a name="line.715"></a> -<span class="sourceLineNo">716</span> initCredentials(job);<a name="line.716"></a> -<span class="sourceLineNo">717</span> }<a name="line.717"></a> -<span class="sourceLineNo">718</span><a name="line.718"></a> -<span class="sourceLineNo">719</span> /**<a name="line.719"></a> -<span class="sourceLineNo">720</span> * Ensures that the given number of reduce tasks for the given job<a name="line.720"></a> -<span class="sourceLineNo">721</span> * configuration does not exceed the number of regions for the given table.<a name="line.721"></a> -<span class="sourceLineNo">722</span> *<a name="line.722"></a> -<span class="sourceLineNo">723</span> * @param table The table to get the region count for.<a name="line.723"></a> -<span class="sourceLineNo">724</span> * @param job The current job to adjust.<a name="line.724"></a> -<span class="sourceLineNo">725</span> * @throws IOException When retrieving the table details fails.<a name="line.725"></a> -<span class="sourceLineNo">726</span> */<a name="line.726"></a> -<span class="sourceLineNo">727</span> public static void limitNumReduceTasks(String table, Job job)<a name="line.727"></a> -<span class="sourceLineNo">728</span> throws IOException {<a name="line.728"></a> -<span class="sourceLineNo">729</span> int regions =<a name="line.729"></a> -<span class="sourceLineNo">730</span> MetaTableAccessor.getRegionCount(job.getConfiguration(), TableName.valueOf(table));<a name="line.730"></a> -<span class="sourceLineNo">731</span> if (job.getNumReduceTasks() > regions)<a name="line.731"></a> -<span class="sourceLineNo">732</span> job.setNumReduceTasks(regions);<a name="line.732"></a> -<span class="sourceLineNo">733</span> }<a name="line.733"></a> -<span class="sourceLineNo">734</span><a name="line.734"></a> -<span class="sourceLineNo">735</span> /**<a name="line.735"></a> -<span class="sourceLineNo">736</span> * Sets the number of reduce tasks for the given job configuration to the<a name="line.736"></a> -<span class="sourceLineNo">737</span> * number of regions the given table has.<a name="line.737"></a> -<span class="sourceLineNo">738</span> *<a name="line.738"></a> -<span class="sourceLineNo">739</span> * @param table The table to get the region count for.<a name="line.739"></a> -<span class="sourceLineNo">740</span> * @param job The current job to adjust.<a name="line.740"></a> -<span class="sourceLineNo">741</span> * @throws IOException When retrieving the table details fails.<a name="line.741"></a> -<span class="sourceLineNo">742</span> */<a name="line.742"></a> -<span class="sourceLineNo">743</span> public static void setNumReduceTasks(String table, Job job)<a name="line.743"></a> -<span class="sourceLineNo">744</span> throws IOException {<a name="line.744"></a> -<span class="sourceLineNo">745</span> job.setNumReduceTasks(MetaTableAccessor.getRegionCount(job.getConfiguration(),<a name="line.745"></a> -<span class="sourceLineNo">746</span> TableName.valueOf(table)));<a name="line.746"></a> -<span class="sourceLineNo">747</span> }<a name="line.747"></a> -<span class="sourceLineNo">748</span><a name="line.748"></a> -<span class="sourceLineNo">749</span> /**<a name="line.749"></a> -<span class="sourceLineNo">750</span> * Sets the number of rows to return and cache with each scanner iteration.<a name="line.750"></a> -<span class="sourceLineNo">751</span> * Higher caching values will enable faster mapreduce jobs at the expense of<a name="line.751"></a> -<span class="sourceLineNo">752</span> * requiring more heap to contain the cached rows.<a name="line.752"></a> -<span class="sourceLineNo">753</span> *<a name="line.753"></a> -<span class="sourceLineNo">754</span> * @param job The current job to adjust.<a name="line.754"></a> -<span class="sourceLineNo">755</span> * @param batchSize The number of rows to return in batch with each scanner<a name="line.755"></a> -<span class="sourceLineNo">756</span> * iteration.<a name="line.756"></a> -<span class="sourceLineNo">757</span> */<a name="line.757"></a> -<span class="sourceLineNo">758</span> public static void setScannerCaching(Job job, int batchSize) {<a name="line.758"></a> -<span class="sourceLineNo">759</span> job.getConfiguration().setInt("hbase.client.scanner.caching", batchSize);<a name="line.759"></a> -<span class="sourceLineNo">760</span> }<a name="line.760"></a> -<span class="sourceLineNo">761</span><a name="line.761"></a> -<span class="sourceLineNo">762</span> /**<a name="line.762"></a> -<span class="sourceLineNo">763</span> * Add HBase and its dependencies (only) to the job configuration.<a name="line.763"></a> -<span class="sourceLineNo">764</span> * <p><a name="line.764"></a> -<span class="sourceLineNo">765</span> * This is intended as a low-level API, facilitating code reuse between this<a name="line.765"></a> -<span class="sourceLineNo">766</span> * class and its mapred counterpart. It also of use to external tools that<a name="line.766"></a> -<span class="sourceLineNo">767</span> * need to build a MapReduce job that interacts with HBase but want<a name="line.767"></a> -<span class="sourceLineNo">768</span> * fine-grained control over the jars shipped to the cluster.<a name="line.768"></a> -<span class="sourceLineNo">769</span> * </p><a name="line.769"></a> -<span class="sourceLineNo">770</span> * @param conf The Configuration object to extend with dependencies.<a name="line.770"></a> -<span class="sourceLineNo">771</span> * @see org.apache.hadoop.hbase.mapred.TableMapReduceUtil<a name="line.771"></a> -<span class="sourceLineNo">772</span> * @see <a href="https://issues.apache.org/jira/browse/PIG-3285">PIG-3285</a><a name="line.772"></a> -<span class="sourceLineNo">773</span> */<a name="line.773"></a> -<span class="sourceLineNo">774</span> public static void addHBaseDependencyJars(Configuration conf) throws IOException {<a name="line.774"></a> -<span class="sourceLineNo">775</span><a name="line.775"></a> -<span class="sourceLineNo">776</span> // PrefixTreeCodec is part of the hbase-prefix-tree module. If not included in MR jobs jar<a name="line.776"></a> -<span class="sourceLineNo">777</span> // dependencies, MR jobs that write encoded hfiles will fail.<a name="line.777"></a> -<span class="sourceLineNo">778</span> // We used reflection here so to prevent a circular module dependency.<a name="line.778"></a> -<span class="sourceLineNo">779</span> // TODO - if we extract the MR into a module, make it depend on hbase-prefix-tree.<a name="line.779"></a> -<span class="sourceLineNo">780</span> Class prefixTreeCodecClass = null;<a name="line.780"></a> -<span class="sourceLineNo">781</span> try {<a name="line.781"></a> -<span class="sourceLineNo">782</span> prefixTreeCodecClass =<a name="line.782"></a> -<span class="sourceLineNo">783</span> Class.forName("org.apache.hadoop.hbase.codec.prefixtree.PrefixTreeCodec");<a name="line.783"></a> -<span class="sourceLineNo">784</span> } catch (ClassNotFoundException e) {<a name="line.784"></a> -<span class="sourceLineNo">785</span> // this will show up in unit tests but should not show in real deployments<a name="line.785"></a> -<span class="sourceLineNo">786</span> LOG.warn("The hbase-prefix-tree module jar containing PrefixTreeCodec is not present." +<a name="line.786"></a> -<span class="sourceLineNo">787</span> " Continuing without it.");<a name="line.787"></a> -<span class="sourceLineNo">788</span> }<a name="line.788"></a> -<span class="sourceLineNo">789</span><a name="line.789"></a> -<span class="sourceLineNo">790</span> addDependencyJarsForClasses(conf,<a name="line.790"></a> -<span class="sourceLineNo">791</span> // explicitly pull a class from each module<a name="line.791"></a> -<span class="sourceLineNo">792</span> org.apache.hadoop.hbase.HConstants.class, // hbase-common<a name="line.792"></a> -<span class="sourceLineNo">793</span> org.apache.hadoop.hbase.protobuf.generated.ClientProtos.class, // hbase-protocol<a name="line.793"></a> -<span class="sourceLineNo">794</span> org.apache.hadoop.hbase.client.Put.class, // hbase-client<a name="line.794"></a> -<span class="sourceLineNo">795</span> org.apache.hadoop.hbase.CompatibilityFactory.class, // hbase-hadoop-compat<a name="line.795"></a> -<span class="sourceLineNo">796</span> org.apache.hadoop.hbase.mapreduce.TableMapper.class, // hbase-server<a name="line.796"></a> -<span class="sourceLineNo">797</span> prefixTreeCodecClass, // hbase-prefix-tree (if null will be skipped)<a name="line.797"></a> -<span class="sourceLineNo">798</span> // pull necessary dependencies<a name="line.798"></a> -<span class="sourceLineNo">799</span> org.apache.zookeeper.ZooKeeper.class,<a name="line.799"></a> -<span class="sourceLineNo">800</span> io.netty.channel.Channel.class,<a name="line.800"></a> -<span class="sourceLineNo">801</span> com.google.protobuf.Message.class,<a name="line.801"></a> -<span class="sourceLineNo">802</span> com.google.common.collect.Lists.class,<a name="line.802"></a> -<span class="sourceLineNo">803</span> org.apache.htrace.Trace.class,<a name="line.803"></a> -<span class="sourceLineNo">804</span> com.codahale.metrics.MetricRegistry.class);<a name="line.804"></a> -<span class="sourceLineNo">805</span> }<a name="line.805"></a> -<span class="sourceLineNo">806</span><a name="line.806"></a> -<span class="sourceLineNo">807</span> /**<a name="line.807"></a> -<span class="sourceLineNo">808</span> * Returns a classpath string built from the content of the "tmpjars" value in {@code conf}.<a name="line.808"></a> -<span class="sourceLineNo">809</span> * Also exposed to shell scripts via `bin/hbase mapredcp`.<a name="line.809"></a> -<span class="sourceLineNo">810</span> */<a name="line.810"></a> -<span class="sourceLineNo">811</span> public static String buildDependencyClasspath(Configuration conf) {<a name="line.811"></a> -<span class="sourceLineNo">812</span> if (conf == null) {<a name="line.812"></a> -<span class="sourceLineNo">813</span> throw new IllegalArgumentException("Must provide a configuration object.");<a name="line.813"></a> -<span class="sourceLineNo">814</span> }<a name="line.814"></a> -<span class="sourceLineNo">815</span> Set<String> paths = new HashSet<String>(conf.getStringCollection("tmpjars"));<a name="line.815"></a> -<span class="sourceLineNo">816</span> if (paths.size() == 0) {<a name="line.816"></a> -<span class="sourceLineNo">817</span> throw new IllegalArgumentException("Configuration contains no tmpjars.");<a name="line.817"></a> -<span class="sourceLineNo">818</span> }<a name="line.818"></a> -<span class="sourceLineNo">819</span> StringBuilder sb = new StringBuilder();<a name="line.819"></a> -<span class="sourceLineNo">820</span> for (String s : paths) {<a name="line.820"></a> -<span class="sourceLineNo">821</span> // entries can take the form 'file:/path/to/file.jar'.<a name="line.821"></a> -<span class="sourceLineNo">822</span> int idx = s.indexOf(":");<a name="line.822"></a> -<span class="sourceLineNo">823</span> if (idx != -1) s = s.substring(idx + 1);<a name="line.823"></a> -<span class="sourceLineNo">824</span> if (sb.length() > 0) sb.append(File.pathSeparator);<a name="line.824"></a> -<span class="sourceLineNo">825</span> sb.append(s);<a name="line.825"></a> -<span class="sourceLineNo">826</span> }<a name="line.826"></a> -<span class="sourceLineNo">827</span> return sb.toString();<a name="line.827"></a> -<span class="sourceLineNo">828</span> }<a name="line.828"></a> -<span class="sourceLineNo">829</span><a name="line.829"></a> -<span class="sourceLineNo">830</span> /**<a name="line.830"></a> -<span class="sourceLineNo">831</span> * Add the HBase dependency jars as well as jars for any of the configured<a name="line.831"></a> -<span class="sourceLineNo">832</span> * job classes to the job configuration, so that JobClient will ship them<a name="line.832"></a> -<span class="sourceLineNo">833</span> * to the cluster and add them to the DistributedCache.<a name="line.833"></a> -<span class="sourceLineNo">834</span> */<a name="line.834"></a> -<span class="sourceLineNo">835</span> public static void addDependencyJars(Job job) throws IOException {<a name="line.835"></a> -<span class="sourceLineNo">836</span> addHBaseDependencyJars(job.getConfiguration());<a name="line.836"></a> -<span class="sourceLineNo">837</span> try {<a name="line.837"></a> -<span class="sourceLineNo">838</span> addDependencyJarsForClasses(job.getConfiguration(),<a name="line.838"></a> -<span class="sourceLineNo">839</span> // when making changes here, consider also mapred.TableMapReduceUtil<a name="line.839"></a> -<span class="sourceLineNo">840</span> // pull job classes<a name="line.840"></a> -<span class="sourceLineNo">841</span> job.getMapOutputKeyClass(),<a name="line.841"></a> -<span class="sourceLineNo">842</span> job.getMapOutputValueClass(),<a name="line.842"></a> -<span class="sourceLineNo">843</span> job.getInputFormatClass(),<a name="line.843"></a> -<span class="sourceLineNo">844</span> job.getOutputKeyClass(),<a name="line.844"></a> -<span class="sourceLineNo">845</span> job.getOutputValueClass(),<a name="line.845"></a> -<span class="sourceLineNo">846</span> job.getOutputFormatClass(),<a name="line.846"></a> -<span class="sourceLineNo">847</span> job.getPartitionerClass(),<a name="line.847"></a> -<span class="sourceLineNo">848</span> job.getCombinerClass());<a name="line.848"></a> -<span class="sourceLineNo">849</span> } catch (ClassNotFoundException e) {<a name="line.849"></a> -<span class="sourceLineNo">850</span> throw new IOException(e);<a name="line.850"></a> -<span class="sourceLineNo">851</span> }<a name="line.851"></a> -<span class="sourceLineNo">852</span> }<a name="line.852"></a> -<span class="sourceLineNo">853</span><a name="line.853"></a> -<span class="sourceLineNo">854</span> /**<a name="line.854"></a> -<span class="sourceLineNo">855</span> * Add the jars containing the given classes to the job's configuration<a name="line.855"></a> -<span class="sourceLineNo">856</span> * such that JobClient will ship them to the cluster and add them to<a name="line.856"></a> -<span class="sourceLineNo">857</span> * the DistributedCache.<a name="line.857"></a> -<span class="sourceLineNo">858</span> * @deprecated rely on {@link #addDependencyJars(Job)} instead.<a name="line.858"></a> -<span class="sourceLineNo">859</span> */<a name="line.859"></a> -<span class="sourceLineNo">860</span> @Deprecated<a name="line.860"></a> -<span class="sourceLineNo">861</span> public static void addDependencyJars(Configuration conf,<a name="line.861"></a> -<span class="sourceLineNo">862</span> Class<?>... classes) throws IOException {<a name="line.862"></a> -<span class="sourceLineNo">863</span> LOG.warn("The addDependencyJars(Configuration, Class<?>...) method has been deprecated since it"<a name="line.863"></a> -<span class="sourceLineNo">864</span> + " is easy to use incorrectly. Most users should rely on addDependencyJars(Job) " +<a name="line.864"></a> -<span class="sourceLineNo">865</span> "instead. See HBASE-8386 for more details.");<a name="line.865"></a> -<span class="sourceLineNo">866</span> addDependencyJarsForClasses(conf, classes);<a name="line.866"></a> -<span class="sourceLineNo">867</span> }<a name="line.867"></a> -<span class="sourceLineNo">868</span><a name="line.868"></a> -<span class="sourceLineNo">869</span> /**<a name="line.869"></a> -<span class="sourceLineNo">870</span> * Add the jars containing the given classes to the job's configuration<a name="line.870"></a> -<span class="sourceLineNo">871</span> * such that JobClient will ship them to the cluster and add them to<a name="line.871"></a> -<span class="sourceLineNo">872</span> * the DistributedCache.<a name="line.872"></a> -<span class="sourceLineNo">873</span> *<a name="line.873"></a> -<span class="sourceLineNo">874</span> * N.B. that this method at most adds one jar per class given. If there is more than one<a name="line.874"></a> -<span class="sourceLineNo">875</span> * jar available containing a class with the same name as a given class, we don't define<a name="line.875"></a> -<span class="sourceLineNo">876</span> * which of those jars might be chosen.<a name="line.876"></a> -<span class="sourceLineNo">877</span> *<a name="line.877"></a> -<span class="sourceLineNo">878</span> * @param conf The Hadoop Configuration to modify<a name="line.878"></a> -<span class="sourceLineNo">879</span> * @param classes will add just those dependencies needed to find the given classes<a name="line.879"></a> -<span class="sourceLineNo">880</span> * @throws IOException if an underlying library call fails.<a name="line.880"></a> -<span class="sourceLineNo">881</span> */<a name="line.881"></a> -<span class="sourceLineNo">882</span> @InterfaceAudience.Private<a name="line.882"></a> -<span class="sourceLineNo">883</span> public static void addDependencyJarsForClasses(Configuration conf,<a name="line.883"></a> -<span class="sourceLineNo">884</span> Class<?>... classes) throws IOException {<a name="line.884"></a> -<span class="sourceLineNo">885</span><a name="line.885"></a> -<span class="sourceLineNo">886</span> FileSystem localFs = FileSystem.getLocal(conf);<a name="line.886"></a> -<span class="sourceLineNo">887</span> Set<String> jars = new HashSet<String>();<a name="line.887"></a> -<span class="sourceLineNo">888</span> // Add jars that are already in the tmpjars variable<a name="line.888"></a> -<span class="sourceLineNo">889</span> jars.addAll(conf.getStringCollection("tmpjars"));<a name="line.889"></a> -<span class="sourceLineNo">890</span><a name="line.890"></a> -<span class="sourceLineNo">891</span> // add jars as we find them to a map of contents jar name so that we can avoid<a name="line.891"></a> -<span class="sourceLineNo">892</span> // creating new jars for classes that have already been packaged.<a name="line.892"></a> -<span class="sourceLineNo">893</span> Map<String, String> packagedClasses = new HashMap<String, String>();<a name="line.893"></a> -<span class="sourceLineNo">894</span><a name="line.894"></a> -<span class="sourceLineNo">895</span> // Add jars containing the specified classes<a name="line.895"></a> -<span class="sourceLineNo">896</span> for (Class<?> clazz : classes) {<a name="line.896"></a> -<span class="sourceLineNo">897</span> if (clazz == null) continue;<a name="line.897"></a> -<span class="sourceLineNo">898</span><a name="line.898"></a> -<span class="sourceLineNo">899</span> Path path = findOrCreateJar(clazz, localFs, packagedClasses);<a name="line.899"></a> -<span class="sourceLineNo">900</span> if (path == null) {<a name="line.900"></a> -<span class="sourceLineNo">901</span> LOG.warn("Could not find jar for class " + clazz +<a name="line.901"></a> -<span class="sourceLineNo">902</span> " in order to ship it to the cluster.");<a name="line.902"></a> -<span class="sourceLineNo">903</span> continue;<a name="line.903"></a> -<span class="sourceLineNo">904</span> }<a name="line.904"></a> -<span class="sourceLineNo">905</span> if (!localFs.exists(path)) {<a name="line.905"></a> -<span class="sourceLineNo">906</span> LOG.warn("Could not validate jar file " + path + " for class "<a name="line.906"></a> -<span class="sourceLineNo">907</span> + clazz);<a name="line.907"></a> -<span class="sourceLineNo">908</span> continue;<a name="line.908"></a> -<span class="sourceLineNo">909</span> }<a name="line.909"></a> -<span class="sourceLineNo">910</span> jars.add(path.toString());<a name="line.910"></a> -<span class="sourceLineNo">911</span> }<a name="line.911"></a> -<span class="sourceLineNo">912</span> if (jars.isEmpty()) return;<a name="line.912"></a> -<span class="sourceLineNo">913</span><a name="line.913"></a> -<span class="sourceLineNo">914</span> conf.set("tmpjars", StringUtils.arrayToString(jars.toArray(new String[jars.size()])));<a name="line.914"></a> -<span class="sourceLineNo">915</span> }<a name="line.915"></a> -<span class="sourceLineNo">916</span><a name="line.916"></a> -<span class="sourceLineNo">917</span> /**<a name="line.917"></a> -<span class="sourceLineNo">918</span> * Finds the Jar for a class or creates it if it doesn't exist. If the class is in<a name="line.918"></a> -<span class="sourceLineNo">919</span> * a directory in the classpath, it creates a Jar on the fly with the<a name="line.919"></a> -<span class="sourceLineNo">920</span> * contents of the directory and returns the path to that Jar. If a Jar is<a name="line.920"></a> -<span class="sourceLineNo">921</span> * created, it is created in the system temporary directory. Otherwise,<a name="line.921"></a> -<span class="sourceLineNo">922</span> * returns an existing jar that contains a class of the same name. Maintains<a name="line.922"></a> -<span class="sourceLineNo">923</span> * a mapping from jar contents to the tmp jar created.<a name="line.923"></a> -<span class="sourceLineNo">924</span> * @param my_class the class to find.<a name="line.924"></a> -<span class="sourceLineNo">925</span> * @param fs the FileSystem with which to qualify the returned path.<a name="line.925"></a> -<span class="sourceLineNo">926</span> * @param packagedClasses a map of class name to path.<a name="line.926"></a> -<span class="sourceLineNo">927</span> * @return a jar file that contains the class.<a name="line.927"></a> -<span class="sourceLineNo">928</span> * @throws IOException<a name="line.928"></a> -<span class="sourceLineNo">929</span> */<a name="line.929"></a> -<span class="sourceLineNo">930</span> private static Path findOrCreateJar(Class<?> my_class, FileSystem fs,<a name="line.930"></a> -<span class="sourceLineNo">931</span> Map<String, String> packagedClasses)<a name="line.931"></a> -<span class="sourceLineNo">932</span> throws IOException {<a name="line.932"></a> -<span class="sourceLineNo">933</span> // attempt to locate an existing jar for the class.<a name="line.933"></a> -<span class="sourceLineNo">934</span> String jar = findContainingJar(my_class, packagedClasses);<a name="line.934"></a> -<span class="sourceLineNo">935</span> if (null == jar || jar.isEmpty()) {<a name="line.935"></a> -<span class="sourceLineNo">936</span> jar = getJar(my_class);<a name="line.936"></a> -<span class="sourceLineNo">937</span> updateMap(jar, packagedClasses);<a name="line.937"></a> -<span class="sourceLineNo">938</span> }<a name="line.938"></a> -<span class="sourceLineNo">939</span><a name="line.939"></a> -<span class="sourceLineNo">940</span> if (null == jar || jar.isEmpty()) {<a name="line.940"></a> -<span class="sourceLineNo">941</span> return null;<a name="line.941"></a> -<span class="sourceLineNo">942</span> }<a name="line.942"></a> -<span class="sourceLineNo">943</span><a name="line.943"></a> -<span class="sourceLineNo">944</span> LOG.debug(String.format("For class %s, using jar %s", my_class.getName(), jar));<a name="line.944"></a> -<span class="sourceLineNo">945</span> return new Path(jar).makeQualified(fs);<a name="line.945"></a> -<span class="sourceLineNo">946</span> }<a name="line.946"></a> -<span class="sourceLineNo">947</span><a name="line.947"></a> -<span class="sourceLineNo">948</span> /**<a name="line.948"></a> -<span class="sourceLineNo">949</span> * Add entries to <code>packagedClasses</code> corresponding to class files<a name="line.949"></a> -<span class="sourceLineNo">950</span> * contained in <code>jar</code>.<a name="line.950"></a> -<span class="sourceLineNo">951</span> * @param jar The jar who's content to list.<a name="line.951"></a> -<span class="sourceLineNo">952</span> * @param packagedClasses map[class -> jar]<a name="line.952"></a> -<span class="sourceLineNo">953</span> */<a name="line.953"></a> -<span class="sourceLineNo">954</span> private static void updateMap(String jar, Map<String, String> packagedClasses) throws IOException {<a name="line.954"></a> -<span class="sourceLineNo">955</span> if (null == jar || jar.isEmpty()) {<a name="line.955"></a> -<span class="sourceLineNo">956</span> return;<a name="line.956"></a> -<span class="sourceLineNo">957</span> }<a name="line.957"></a> -<span class="sourceLineNo">958</span> ZipFile zip = null;<a name="line.958"></a> -<span class="sourceLineNo">959</span> try {<a name="line.959"></a> -<span class="sourceLineNo">960</span> zip = new ZipFile(jar);<a name="line.960"></a> -<span class="sourceLineNo">961</span> for (Enumeration<? extends ZipEntry> iter = zip.entries(); iter.hasMoreElements();) {<a name="line.961"></a> -<span class="sourceLineNo">962</span> ZipEntry entry = iter.nextElement();<a name="line.962"></a> -<span class="sourceLineNo">963</span> if (entry.getName().endsWith("class")) {<a name="line.963"></a> -<span class="sourceLineNo">964</span> packagedClasses.put(entry.getName(), jar);<a name="line.964"></a> -<span class="sourceLineNo">965</span> }<a name="line.965"></a> -<span class="sourceLineNo">966</span> }<a name="line.966"></a> -<span class="sourceLineNo">967</span> } finally {<a name="line.967"></a> -<span class="sourceLineNo">968</span> if (null != zip) zip.close();<a name="line.968"></a> -<span class="sourceLineNo">969</span> }<a name="line.969"></a> -<span class="sourceLineNo">970</span> }<a name="line.970"></a> -<span class="sourceLineNo">971</span><a name="line.971"></a> -<span class="sourceLineNo">972</span> /**<a name="line.972"></a> -<span class="sourceLineNo">973</span> * Find a jar that contains a class of the same name, if any. It will return<a name="line.973"></a> -<span class="sourceLineNo">974</span> * a jar file, even if that is not the first thing on the class path that<a name="line.974"></a> -<span class="sourceLineNo">975</span> * has a class with the same name. Looks first on the classpath and then in<a name="line.975"></a> -<span class="sourceLineNo">976</span> * the <code>packagedClasses</code> map.<a name="line.976"></a> -<span class="sourceLineNo">977</span> * @param my_class the class to find.<a name="line.977"></a> -<span class="sourceLineNo">978</span> * @return a jar file that contains the class, or null.<a name="line.978"></a> -<span class="sourceLineNo">979</span> * @throws IOException<a name="line.979"></a> -<span class="sourceLineNo">980</span> */<a name="line.980"></a> -<span class="sourceLineNo">981</span> private static String findContainingJar(Class<?> my_class, Map<String, String> packagedClasses)<a name="line.981"></a> -<span class="sourceLineNo">982</span> throws IOException {<a name="line.982"></a> -<span class="sourceLineNo">983</span> ClassLoader loader = my_class.getClassLoader();<a name="line.983"></a> -<span class="sourceLineNo">984</span><a name="line.984"></a> -<span class="sourceLineNo">985</span> String class_file = my_class.getName().replaceAll("\\.", "/") + ".class";<a name="line.985"></a> -<span class="sourceLineNo">986</span><a name="line.986"></a> -<span class="sourceLineNo">987</span> if (loader != null) {<a name="line.987"></a> -<span class="sourceLineNo">988</span> // first search the classpath<a name="line.988"></a> -<span class="sourceLineNo">989</span> for (Enumeration<URL> itr = loader.getResources(class_file); itr.hasMoreElements();) {<a name="line.989"></a> -<span class="sourceLineNo">990</span> URL url = itr.nextElement();<a name="line.990"></a> -<span class="sourceLineNo">991</span> if ("jar".equals(url.getProtocol())) {<a name="line.991"></a> -<span class="sourceLineNo">992</span> String toReturn = url.getPath();<a name="line.992"></a> -<span class="sourceLineNo">993</span> if (toReturn.startsWith("file:")) {<a name="line.993"></a> -<span class="sourceLineNo">994</span> toReturn = toReturn.substring("file:".length());<a name="line.994"></a> -<span class="sourceLineNo">995</span> }<a name="line.995"></a> -<span class="sourceLineNo">996</span> // URLDecoder is a misnamed class, since it actually decodes<a name="line.996"></a> -<span class="sourceLineNo">997</span> // x-www-form-urlencoded MIME type rather than actual<a name="line.997"></a> -<span class="sourceLineNo">998</span> // URL encoding (which the file path has). Therefore it would<a name="line.998"></a> -<span class="sourceLineNo">999</span> // decode +s to ' 's which is incorrect (spaces are actually<a name="line.999"></a> -<span class="sourceLineNo">1000</span> // either unencoded or encoded as "%20"). Replace +s first, so<a name="line.1000"></a> -<span class="sourceLineNo">1001</span> // that they are kept sacred during the decoding process.<a name="line.1001"></a> -<span class="sourceLineNo">1002</span> toReturn = toReturn.replaceAll("\\+", "%2B");<a name="line.1002"></a> -<span class="sourceLineNo">1003</span> toReturn = URLDecoder.decode(toReturn, "UTF-8");<a name="line.1003"></a> -<span class="sourceLineNo">1004</span> return toReturn.replaceAll("!.*$", "");<a name="line.1004"></a> -<span class="sourceLineNo">1005</span> }<a name="line.1005"></a> -<span class="sourceLineNo">1006</span> }<a name="line.1006"></a> -<span class="sourceLineNo">1007</span> }<a name="line.1007"></a> -<span class="sourceLineNo">1008</span><a name="line.1008"></a> -<span class="sourceLineNo">1009</span> // now look in any jars we've packaged using JarFinder. Returns null when<a name="line.1009"></a> -<span class="sourceLineNo">1010</span> // no jar is found.<a name="line.1010"></a> -<span class="sourceLineNo">1011</span> return packagedClasses.get(class_file);<a name="line.1011"></a> -<span class="sourceLineNo">1012</span> }<a name="line.1012"></a> -<span class="sourceLineNo">1013</span><a name="line.1013"></a> -<span class="sourceLineNo">1014</span> /**<a name="line.1014"></a> -<span class="sourceLineNo">1015</span> * Invoke 'getJar' on a custom JarFinder implementation. Useful for some job<a name="line.1015"></a> -<span class="sourceLineNo">1016</span> * configuration contexts (HBASE-8140) and also for testing on MRv2.<a name="line.1016"></a> -<span class="sourceLineNo">1017</span> * check if we have HADOOP-9426.<a name="line.1017"></a> -<span class="sourceLineNo">1018</span> * @param my_class the class to find.<a name="line.1018"></a> -<span class="sourceLineNo">1019</span> * @return a jar file that contains the class, or null.<a name="line.1019"></a> -<span class="sourceLineNo">1020</span> */<a name="line.1020"></a> -<span class="sourceLineNo">1021</span> private static String getJar(Class<?> my_class) {<a name="line.1021"></a> -<span class="sourceLineNo">1022</span> String ret = null;<a name="line.1022"></a> -<span class="sourceLineNo">1023</span> try {<a name="line.1023"></a> -<span class="sourceLineNo">1024</span> ret = JarFinder.getJar(my_class);<a name="line.1024"></a> -<span class="sourceLineNo">1025</span> } catch (Exception e) {<a name="line.1025"></a> -<span class="sourceLineNo">1026</span> // toss all other exceptions, related to reflection failure<a name="line.1026"></a> -<span class="sourceLineNo">1027</span> throw new RuntimeException("getJar invocation failed.", e);<a name="line.1027"></a> -<span class="sourceLineNo">1028</span> }<a name="line.1028"></a> -<span class="sourceLineNo">1029</span><a name="line.1029"></a> -<span class="sourceLineNo">1030</span> return ret;<a name="line.1030"></a> -<span class="sourceLineNo">1031</span> }<a name="line.1031"></a> -<span class="sourceLineNo">1032</span>}<a name="line.1032"></a> +<span class="sourceLineNo">712</span> /**<a name="line.712"></a> +<span class="sourceLineNo">713</span> * Ensures that the given number of reduce tasks for the given job<a name="line.713"></a> +<span class="sourceLineNo">714</span> * configuration does not exceed the number of regions for the given table.<a name="line.714"></a> +<span class="sourceLineNo">715</span> *<a name="line.715"></a> +<span class="sourceLineNo">716</span> * @param table The table to get the region count for.<a name="line.716"></a> +<span class="sourceLineNo">717</span> * @param job The current job to adjust.<a name="line.717"></a> +<span class="sourceLineNo">718</span> * @throws IOException When retrieving the table details fails.<a name="line.718"></a> +<span class="sourceLineNo">719</span> */<a name="line.719"></a> +<span class="sourceLineNo">720</span> public static void limitNumReduceTasks(String table, Job job)<a name="line.720"></a> +<span class="sourceLineNo">721</span> throws IOException {<a name="line.721"></a> +<span class="sourceLineNo">722</span> int regions =<a name="line.722"></a> +<span class="sourceLineNo">723</span> MetaTableAccessor.getRegionCount(job.getConfiguration(), TableName.valueOf(table));<a name="line.723"></a> +<span class="sourceLineNo">724</span> if (job.getNumReduceTasks() > regions)<a name="line.724"></a> +<span class="sourceLineNo">725</span> job.setNumReduceTasks(regions);<a name="line.725"></a> +<span class="sourceLineNo">726</span> }<a name="line.726"></a> +<span class="sourceLineNo">727</span><a name="line.727"></a> +<span class="sourceLineNo">728</span> /**<a name="line.728"></a> +<span class="sourceLineNo">729</span> * Sets the number of reduce tasks for the given job configuration to the<a name="line.729"></a> +<span class="sourceLineNo">730</span> * number of regions the given table has.<a name="line.730"></a> +<span class="sourceLineNo">731</span> *<a name="line.731"></a> +<span class="sourceLineNo">732</span> * @param table The table to get the region count for.<a name="line.732"></a> +<span class="sourceLineNo">733</span> * @param job The current job to adjust.<a name="line.733"></a> +<span class="sourceLineNo">734</span> * @throws IOException When retrieving the table details fails.<a name="line.734"></a> +<span class="sourceLineNo">735</span> */<a name="line.735"></a> +<span class="sourceLineNo">736</span> public static void setNumReduceTasks(String table, Job job)<a name="line.736"></a> +<span class="sourceLineNo">737</span> throws IOException {<a name="line.737"></a> +<span class="sourceLineNo">738</span> job.setNumReduceTasks(MetaTableAccessor.getRegionCount(job.getConfiguration(),<a name="line.738"></a> +<span class="sourceLineNo">739</span> TableName.valueOf(table)));<a name="line.739"></a> +<span class="sourceLineNo">740</span> }<a name="line.740"></a> +<span class="sourceLineNo">741</span><a name="line.741"></a> +<span class="sourceLineNo">742</span> /**<a name="line.742"></a> +<span class="sourceLineNo">743</span> * Sets the number of rows to return and cache with each scanner iteration.<a name="line.743"></a> +<span class="sourceLineNo">744</span> * Higher caching values will enable faster mapreduce jobs at the expense of<a name="line.744"></a> +<span class="sourceLineNo">745</span> * requiring more heap to contain the cached rows.<a name="line.745"></a> +<span class="sourceLineNo">746</span> *<a name="line.746"></a> +<span class="sourceLineNo">747</span> * @param job The current job to adjust.<a name="line.747"></a> +<span class="sourceLineNo">748</span> * @param batchSize The number of rows to return in batch with each scanner<a name="line.748"></a> +<span class="sourceLineNo">749</span> * iteration.<a name="line.749"></a> +<span class="sourceLineNo">750</span> */<a name="line.750"></a> +<span class="sourceLineNo">751</span> public static void setScannerCaching(Job job, int batchSize) {<a name="line.751"></a> +<span class="sourceLineNo">752</span> job.getConfiguration().setInt("hbase.client.scanner.caching", batchSize);<a name="line.752"></a> +<span class="sourceLineNo">753</span> }<a name="line.753"></a> +<span class="sourceLineNo">754</span><a name="line.754"></a> +<span class="sourceLineNo">755</span> /**<a name="line.755"></a> +<span class="sourceLineNo">756</span> * Add HBase and its dependencies (only) to the job configuration.<a name="line.756"></a> +<span class="sourceLineNo">757</span> * <p><a name="line.757"></a> +<span class="sourceLineNo">758</span> * This is intended as a low-level API, facilitating code reuse between this<a name="line.758"></a> +<span class="sourceLineNo">759</span> * class and its mapred counterpart. It also of use to external tools that<a name="line.759"></a> +<span class="sourceLineNo">760</span> * need to build a MapReduce job that interacts with HBase but want<a name="line.760"></a> +<span class="sourceLineNo">761</span> * fine-grained control over the jars shipped to the cluster.<a name="line.761"></a> +<span class="sourceLineNo">762</span> * </p><a name="line.762"></a> +<span class="sourceLineNo">763</span> * @param conf The Configuration object to extend with dependencies.<a name="line.763"></a> +<span class="sourceLineNo">764</span> * @see org.apache.hadoop.hbase.mapred.TableMapReduceUtil<a name="line.764"></a> +<span class="sourceLineNo">765</span> * @see <a href="https://issues.apache.org/jira/browse/PIG-3285">PIG-3285</a><a name="line.765"></a> +<span class="sourceLineNo">766</span> */<a name="line.766"></a> +<span class="sourceLineNo">767</span> public static void addHBaseDependencyJars(Configuration conf) throws IOException {<a name="line.767"></a> +<span class="sourceLineNo">768</span><a name="line.768"></a> +<span class="sourceLineNo">769</span> // PrefixTreeCodec is part of the hbase-prefix-tree module. If not included in MR jobs jar<a name="line.769"></a> +<span class="sourceLineNo">770</span> // dependencies, MR jobs that write encoded hfiles will fail.<a name="line.770"></a> +<span class="sourceLineNo">771</span> // We used reflection here so to prevent a circular module dependency.<a name="line.771"></a> +<span class="sourceLineNo">772</span> // TODO - if we extract the MR into a module, make it depend on hbase-prefix-tree.<a name="line.772"></a> +<span class="sourceLineNo">773</span> Class prefixTreeCodecClass = null;<a name="line.773"></a> +<span class="sourceLineNo">774</span> try {<a name="line.774"></a> +<span class="sourceLineNo">775</span> prefixTreeCodecClass =<a name="line.775"></a> +<span class="sourceLineNo">776</span> Class.forName("org.apache.hadoop.hbase.codec.prefixtree.PrefixTreeCodec");<a name="line.776"></a> +<span class="sourceLineNo">777</span> } catch (ClassNotFoundException e) {<a name="line.777"></a> +<span class="sourceLineNo">778</span> // this will show up in unit tests but should not show in real deployments<a name="line.778"></a> +<span class="sourceLineNo">779</span> LOG.warn("The hbase-prefix-tree module jar containing PrefixTreeCodec is not present." +<a name="line.779"></a> +<span class="sourceLineNo">780</span> " Continuing without it.");<a name="line.780"></a> +<span class="sourceLineNo">781</span> }<a name="line.781"></a> +<span class="sourceLineNo">782</span><a name="line.782"></a> +<span class="sourceLineNo">783</span> addDependencyJarsForClasses(conf,<a name="line.783"></a> +<span class="sourceLineNo">784</span> // explicitly pull a class from each module<a name="line.784"></a> +<span class="sourceLineNo">785</span> org.apache.hadoop.hbase.HConstants.class, // hbase-common<a name="line.785"></a> +<span class="sourceLineNo">786</span> org.apache.hadoop.hbase.protobuf.generated.ClientProtos.class, // hbase-protocol<a name="line.786"></a> +<span class="sourceLineNo">787</span> org.apache.hadoop.hbase.client.Put.class, // hbase-client<a name="line.787"></a> +<span class="sourceLineNo">788</span> org.apache.hadoop.hbase.CompatibilityFactory.class, // hbase-hadoop-compat<a name="line.788"></a> +<span class="sourceLineNo">789</span> org.apache.hadoop.hbase.mapreduce.TableMapper.class, // hbase-server<a name="line.789"></a> +<span class="sourceLineNo">790</span> prefixTreeCodecClass, // hbase-prefix-tree (if null will be skipped)<a name="line.790"></a> +<span class="sourceLineNo">791</span> // pull necessary dependencies<a name="line.791"></a> +<span class="sourceLineNo">792</span> org.apache.zookeeper.ZooKeeper.class,<a name="line.792"></a> +<span class="sourceLineNo">793</span> io.netty.channel.Channel.class,<a name="line.793"></a> +<span class="sourceLineNo">794</span> com.google.protobuf.Message.class,<a name="line.794"></a> +<span class="sourceLineNo">795</span> com.google.common.collect.Lists.class,<a name="line.795"></a> +<span class="sourceLineNo">796</span> org.apache.htrace.Trace.class,<a name="line.796"></a> +<span class="sourceLineNo">797</span> com.codahale.metrics.MetricRegistry.class);<a name="line.797"></a> +<span class="sourceLineNo">798</span> }<a name="line.798"></a> +<span class="sourceLineNo">799</span><a name="line.799"></a> +<span class="sourceLineNo">800</span> /**<a name="line.800"></a> +<span class="sourceLineNo">801</span> * Returns a classpath string built from the content of the "tmpjars" value in {@code conf}.<a name="line.801"></a> +<span class="sourceLineNo">802</span> * Also exposed to shell scripts via `bin/hbase mapredcp`.<a name="line.802"></a> +<span class="sourceLineNo">803</span> */<a name="line.803"></a> +<span class="sourceLineNo">804</span> public static String buildDependencyClasspath(Configuration conf) {<a name="line.804"></a> +<span class="sourceLineNo">805</span> if (conf == null) {<a name="line.805"></a> +<span class="sourceLineNo">806</span> throw new IllegalArgumentException("Must provide a configuration object.");<a name="line.806"></a> +<span class="sourceLineNo">807</span> }<a name="line.807"></a> +<span class="sourceLineNo">808</span> Set<String> paths = new HashSet<String>(conf.getStringCollection("tmpjars"));<a name="line.808"></a> +<span class="sourceLineNo">809</span> if (paths.size() == 0) {<a name="line.809"></a> +<span class="sourceLineNo">810</span> throw new IllegalArgumentException("Configuration contains no tmpjars.");<a name="line.810"></a> +<span class="sourceLineNo">811</span> }<a name="line.811"></a> +<span class="sourceLineNo">812</span> StringBuilder sb = new StringBuilder();<a name="line.812"></a> +<span class="sourceLineNo">813</span> for (String s : paths) {<a name="line.813"></a> +<span class="sourceLineNo">814</span> // entries can take the form 'file:/path/to/file.jar'.<a name="line.814"></a> +<span class="sourceLineNo">815</span> int idx = s.indexOf(":");<a name="line.815"></a> +<span class="sourceLineNo">816</span> if (idx != -1) s = s.substring(idx + 1);<a name="line.816"></a> +<span class="sourceLineNo">817</span> if (sb.length() > 0) sb.append(File.pathSeparator);<a name="line.817"></a> +<span class="sourceLineNo">818</span> sb.append(s);<a name="line.818"></a> +<span class="sourceLineNo">819</span> }<a name="line.819"></a> +<span class="sourceLineNo">820</span> return sb.toString();<a name="line.820"></a> +<span class="sourceLineNo">821</span> }<a name="line.821"></a> +<span class="sourceLineNo">822</span><a name="line.822"></a> +<span class="sourceLineNo">823</span> /**<a name="line.823"></a> +<span class="sourceLineNo">824</span> * Add the HBase dependency jars as well as jars for any of the configured<a name="line.824"></a> +<span class="sourceLineNo">825</span> * job classes to the job configuration, so that JobClient will ship them<a name="line.825"></a> +<span class="sourceLineNo">826</span> * to the cluster and add them to the DistributedCache.<a name="line.826"></a> +<span class="sourceLineNo">827</span> */<a name="line.827"></a> +<span class="sourceLineNo">828</span> public static void addDependencyJars(Job job) throws IOException {<a name="line.828"></a> +<span class="sourceLineNo">829</span> addHBaseDependencyJars(job.getConfiguration());<a name="line.829"></a> +<span class="sourceLineNo">830</span> try {<a name="line.830"></a> +<span class="sourceLineNo">831</span> addDependencyJarsForClasses(job.getConfiguration(),<a name="line.831"></a> +<span class="sourceLineNo">832</span> // when making changes here, consider also mapred.TableMapReduceUtil<a name="line.832"></a> +<span class="sourceLineNo">833</span> // pull job classes<a name="line.833"></a> +<span class="sourceLineNo">834</span> job.getMapOutputKeyClass(),<a name="line.834"></a> +<span class="sourceLineNo">835</span> job.getMapOutputValueClass(),<a name="line.835"></a> +<span class="sourceLineNo">836</span> job.getInputFormatClass(),<a name="line.836"></a> +<span class="sourceLineNo">837</span> job.getOutputKeyClass(),<a name="line.837"></a> +<span class="sourceLineNo">838</span> job.getOutputValueClass(),<a name="line.838"></a> +<span class="sourceLineNo">839</span> job.getOutputFormatClass(),<a name="line.839"></a> +<span class="sourceLineNo">840</span> job.getPartitionerClass(),<a name="line.840"></a> +<span class="sourceLineNo">841</span> job.getCombinerClass());<a name="line.841"></a> +<span class="sourceLineNo">842</span> } catch (ClassNotFoundException e) {<a name="line.842"></a> +<span class="sourceLineNo">843</span> throw new IOException(e);<a name="line.843"></a> +<span class="sourceLineNo">844</span> }<a name="line.844"></a> +<span class="sourceLineNo">845</span> }<a name="line.845"></a> +<span class="sourceLineNo">846</span><a name="line.846"></a> +<span class="sourceLineNo">847</span> /**<a name="line.847"></a> +<span class="sourceLineNo">848</span> * Add the jars containing the given classes to the job's configuration<a name="line.848"></a> +<span class="sourceLineNo">849</span> * such that JobClient will ship them to the cluster and add them to<a name="line.849"></a> +<span class="sourceLineNo">850</span> * the DistributedCache.<a name="line.850"></a> +<span class="sourceLineNo">851</span> * @deprecated rely on {@link #addDependencyJars(Job)} instead.<a name="line.851"></a> +<span class="sourceLineNo">852</span> */<a name="line.852"></a> +<span class="sourceLineNo">853</span> @Deprecated<a name="line.853"></a> +<span class="sourceLineNo">854</span> public static void addDependencyJars(Configuration conf,<a name="line.854"></a> +<span class="sourceLineNo">855</span> Class<?>... classes) throws IOException {<a name="line.855"></a> +<span class="sourceLineNo">856</span> LOG.warn("The addDependencyJars(Configuration, Class<?>...) method has been deprecated since it"<a name="line.856"></a> +<span class="sourceLineNo">857</span> + " is easy to use incorrectly. Most users should rely on addDependencyJars(Job) " +<a name="line.857"></a> +<span class="sourceLineNo">858</span> "instead. See HBASE-8386 for more details.");<a name="line.858"></a> +<span class="sourceLineNo">859</span> addDependencyJarsForClasses(conf, classes);<a name="line.859"></a> +<span class="sourceLineNo">860</span> }<a name="line.860"></a> +<span class="sourceLineNo">861</span><a name="line.861"></a> +<span class="sourceLineNo">862</span> /**<a name="line.862"></a> +<span class="sourceLineNo">863</span> * Add the jars containing the given classes to the job's configuration<a name="line.863"></a> +<span class="sourceLineNo">864</span> * such that JobClient will ship them to the cluster and add them to<a name="line.864"></a> +<span class="sourceLineNo">865</span> * the DistributedCache.<a name="line.865"></a> +<span class="sourceLineNo">866</span> *<a name="line.866"></a> +<span class="sourceLineNo">867</span> * N.B. that this method at most adds one jar per class given. If there is more than one<a name="line.867"></a> +<span class="sourceLineNo">868</span> * jar available containing a class with the same name as a given class, we don't define<a name="line.868"></a> +<span class="sourceLineNo">869</span> * which of those jars might be chosen.<a name="line.869"></a> +<span class="sourceLineNo">870</span> *<a name="line.870"></a> +<span class="sourceLineNo">871</span> * @param conf The Hadoop Configuration to modify<a name="line.871"></a> +<span class="sourceLineNo">872</span> * @param classes will add just those dependencies needed to find the given classes<a name="line.872"></a> +<span class="sourceLineNo">873</span> * @throws IOException if an underlying library call fails.<a name="line.873"></a> +<span class="sourceLineNo">874</span> */<a name="line.874"></a> +<span class="sourceLineNo">875</span> @InterfaceAudience.Private<a name="line.875"></a> +<span class="sourceLineNo">876</span> public static void addDependencyJarsForClasses(Configuration conf,<a name="line.876"></a> +<span class="sourceLineNo">877</span> Class<?>... classes) throws IOException {<a name="line.877"></a> +<span class="
<TRUNCATED>
