blob: 00229eefce81c646fa58a51ce5ff2740e25cd31b [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/*<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.tool;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.nio.ByteBuffer;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.util.List;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.util.Map;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.conf.Configuration;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.fs.Path;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.TableName;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.TableNotFoundException;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.29"></a>
<span class="sourceLineNo">030</span><a name="line.30"></a>
<span class="sourceLineNo">031</span>/**<a name="line.31"></a>
<span class="sourceLineNo">032</span> * The tool to let you load the output of {@code HFileOutputFormat} into an existing table<a name="line.32"></a>
<span class="sourceLineNo">033</span> * programmatically. Not thread safe.<a name="line.33"></a>
<span class="sourceLineNo">034</span> */<a name="line.34"></a>
<span class="sourceLineNo">035</span>@InterfaceAudience.Public<a name="line.35"></a>
<span class="sourceLineNo">036</span>public interface BulkLoadHFiles {<a name="line.36"></a>
<span class="sourceLineNo">037</span><a name="line.37"></a>
<span class="sourceLineNo">038</span> static final String RETRY_ON_IO_EXCEPTION = "hbase.bulkload.retries.retryOnIOException";<a name="line.38"></a>
<span class="sourceLineNo">039</span> static final String MAX_FILES_PER_REGION_PER_FAMILY =<a name="line.39"></a>
<span class="sourceLineNo">040</span> "hbase.mapreduce.bulkload.max.hfiles.perRegion.perFamily";<a name="line.40"></a>
<span class="sourceLineNo">041</span> static final String ASSIGN_SEQ_IDS = "hbase.mapreduce.bulkload.assign.sequenceNumbers";<a name="line.41"></a>
<span class="sourceLineNo">042</span> static final String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.42"></a>
<span class="sourceLineNo">043</span> static final String IGNORE_UNMATCHED_CF_CONF_KEY = "ignore.unmatched.families";<a name="line.43"></a>
<span class="sourceLineNo">044</span> static final String ALWAYS_COPY_FILES = "always.copy.files";<a name="line.44"></a>
<span class="sourceLineNo">045</span><a name="line.45"></a>
<span class="sourceLineNo">046</span> /**<a name="line.46"></a>
<span class="sourceLineNo">047</span> * Represents an HFile waiting to be loaded. An queue is used in this class in order to support<a name="line.47"></a>
<span class="sourceLineNo">048</span> * the case where a region has split during the process of the load. When this happens, the HFile<a name="line.48"></a>
<span class="sourceLineNo">049</span> * is split into two physical parts across the new region boundary, and each part is added back<a name="line.49"></a>
<span class="sourceLineNo">050</span> * into the queue. The import process finishes when the queue is empty.<a name="line.50"></a>
<span class="sourceLineNo">051</span> */<a name="line.51"></a>
<span class="sourceLineNo">052</span> @InterfaceAudience.Public<a name="line.52"></a>
<span class="sourceLineNo">053</span> public static class LoadQueueItem {<a name="line.53"></a>
<span class="sourceLineNo">054</span><a name="line.54"></a>
<span class="sourceLineNo">055</span> private final byte[] family;<a name="line.55"></a>
<span class="sourceLineNo">056</span><a name="line.56"></a>
<span class="sourceLineNo">057</span> private final Path hfilePath;<a name="line.57"></a>
<span class="sourceLineNo">058</span><a name="line.58"></a>
<span class="sourceLineNo">059</span> public LoadQueueItem(byte[] family, Path hfilePath) {<a name="line.59"></a>
<span class="sourceLineNo">060</span> this.family = family;<a name="line.60"></a>
<span class="sourceLineNo">061</span> this.hfilePath = hfilePath;<a name="line.61"></a>
<span class="sourceLineNo">062</span> }<a name="line.62"></a>
<span class="sourceLineNo">063</span><a name="line.63"></a>
<span class="sourceLineNo">064</span> @Override<a name="line.64"></a>
<span class="sourceLineNo">065</span> public String toString() {<a name="line.65"></a>
<span class="sourceLineNo">066</span> return "family:" + Bytes.toString(family) + " path:" + hfilePath.toString();<a name="line.66"></a>
<span class="sourceLineNo">067</span> }<a name="line.67"></a>
<span class="sourceLineNo">068</span><a name="line.68"></a>
<span class="sourceLineNo">069</span> public byte[] getFamily() {<a name="line.69"></a>
<span class="sourceLineNo">070</span> return family;<a name="line.70"></a>
<span class="sourceLineNo">071</span> }<a name="line.71"></a>
<span class="sourceLineNo">072</span><a name="line.72"></a>
<span class="sourceLineNo">073</span> public Path getFilePath() {<a name="line.73"></a>
<span class="sourceLineNo">074</span> return hfilePath;<a name="line.74"></a>
<span class="sourceLineNo">075</span> }<a name="line.75"></a>
<span class="sourceLineNo">076</span> }<a name="line.76"></a>
<span class="sourceLineNo">077</span><a name="line.77"></a>
<span class="sourceLineNo">078</span> /**<a name="line.78"></a>
<span class="sourceLineNo">079</span> * Perform a bulk load of the given directory into the given pre-existing table.<a name="line.79"></a>
<span class="sourceLineNo">080</span> * @param tableName the table to load into<a name="line.80"></a>
<span class="sourceLineNo">081</span> * @param family2Files map of family to List of hfiles<a name="line.81"></a>
<span class="sourceLineNo">082</span> * @throws TableNotFoundException if table does not yet exist<a name="line.82"></a>
<span class="sourceLineNo">083</span> */<a name="line.83"></a>
<span class="sourceLineNo">084</span> Map&lt;LoadQueueItem, ByteBuffer&gt; bulkLoad(TableName tableName, Map&lt;byte[], List&lt;Path&gt;&gt; family2Files)<a name="line.84"></a>
<span class="sourceLineNo">085</span> throws TableNotFoundException, IOException;<a name="line.85"></a>
<span class="sourceLineNo">086</span><a name="line.86"></a>
<span class="sourceLineNo">087</span> /**<a name="line.87"></a>
<span class="sourceLineNo">088</span> * Disables replication for all bulkloads done via this instance, when bulkload replication is<a name="line.88"></a>
<span class="sourceLineNo">089</span> * configured.<a name="line.89"></a>
<span class="sourceLineNo">090</span> */<a name="line.90"></a>
<span class="sourceLineNo">091</span> void disableReplication();<a name="line.91"></a>
<span class="sourceLineNo">092</span><a name="line.92"></a>
<span class="sourceLineNo">093</span> /** Returns true if replication has been disabled. */<a name="line.93"></a>
<span class="sourceLineNo">094</span> boolean isReplicationDisabled();<a name="line.94"></a>
<span class="sourceLineNo">095</span><a name="line.95"></a>
<span class="sourceLineNo">096</span> /**<a name="line.96"></a>
<span class="sourceLineNo">097</span> * Perform a bulk load of the given directory into the given pre-existing table.<a name="line.97"></a>
<span class="sourceLineNo">098</span> * @param tableName the table to load into<a name="line.98"></a>
<span class="sourceLineNo">099</span> * @param dir the directory that was provided as the output path of a job using<a name="line.99"></a>
<span class="sourceLineNo">100</span> * {@code HFileOutputFormat}<a name="line.100"></a>
<span class="sourceLineNo">101</span> * @throws TableNotFoundException if table does not yet exist<a name="line.101"></a>
<span class="sourceLineNo">102</span> */<a name="line.102"></a>
<span class="sourceLineNo">103</span> Map&lt;LoadQueueItem, ByteBuffer&gt; bulkLoad(TableName tableName, Path dir)<a name="line.103"></a>
<span class="sourceLineNo">104</span> throws TableNotFoundException, IOException;<a name="line.104"></a>
<span class="sourceLineNo">105</span><a name="line.105"></a>
<span class="sourceLineNo">106</span> static BulkLoadHFiles create(Configuration conf) {<a name="line.106"></a>
<span class="sourceLineNo">107</span> return new BulkLoadHFilesTool(conf);<a name="line.107"></a>
<span class="sourceLineNo">108</span> }<a name="line.108"></a>
<span class="sourceLineNo">109</span><a name="line.109"></a>
<span class="sourceLineNo">110</span>}<a name="line.110"></a>
</pre>
</div>
</body>
</html>