blob: bd691f2258fbe9a61721315f464c689c13401569 [file] [log] [blame]
<!DOCTYPE HTML>
<html lang="en">
<head>
<!-- Generated by javadoc (17) -->
<title>Source code</title>
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="source: package: org.apache.hadoop.hbase.regionserver, class: StoreUtils">
<meta name="generator" content="javadoc/SourceToHTMLConverter">
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body class="source-page">
<main role="main">
<div class="source-container">
<pre><span class="source-line-no">001</span><span id="line-1">/*</span>
<span class="source-line-no">002</span><span id="line-2"> * Licensed to the Apache Software Foundation (ASF) under one</span>
<span class="source-line-no">003</span><span id="line-3"> * or more contributor license agreements. See the NOTICE file</span>
<span class="source-line-no">004</span><span id="line-4"> * distributed with this work for additional information</span>
<span class="source-line-no">005</span><span id="line-5"> * regarding copyright ownership. The ASF licenses this file</span>
<span class="source-line-no">006</span><span id="line-6"> * to you under the Apache License, Version 2.0 (the</span>
<span class="source-line-no">007</span><span id="line-7"> * "License"); you may not use this file except in compliance</span>
<span class="source-line-no">008</span><span id="line-8"> * with the License. You may obtain a copy of the License at</span>
<span class="source-line-no">009</span><span id="line-9"> *</span>
<span class="source-line-no">010</span><span id="line-10"> * http://www.apache.org/licenses/LICENSE-2.0</span>
<span class="source-line-no">011</span><span id="line-11"> *</span>
<span class="source-line-no">012</span><span id="line-12"> * Unless required by applicable law or agreed to in writing, software</span>
<span class="source-line-no">013</span><span id="line-13"> * distributed under the License is distributed on an "AS IS" BASIS,</span>
<span class="source-line-no">014</span><span id="line-14"> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
<span class="source-line-no">015</span><span id="line-15"> * See the License for the specific language governing permissions and</span>
<span class="source-line-no">016</span><span id="line-16"> * limitations under the License.</span>
<span class="source-line-no">017</span><span id="line-17"> */</span>
<span class="source-line-no">018</span><span id="line-18">package org.apache.hadoop.hbase.regionserver;</span>
<span class="source-line-no">019</span><span id="line-19"></span>
<span class="source-line-no">020</span><span id="line-20">import java.io.IOException;</span>
<span class="source-line-no">021</span><span id="line-21">import java.util.Collection;</span>
<span class="source-line-no">022</span><span id="line-22">import java.util.List;</span>
<span class="source-line-no">023</span><span id="line-23">import java.util.Optional;</span>
<span class="source-line-no">024</span><span id="line-24">import java.util.OptionalInt;</span>
<span class="source-line-no">025</span><span id="line-25">import java.util.OptionalLong;</span>
<span class="source-line-no">026</span><span id="line-26">import java.util.function.Predicate;</span>
<span class="source-line-no">027</span><span id="line-27">import java.util.function.ToLongFunction;</span>
<span class="source-line-no">028</span><span id="line-28">import java.util.stream.Collectors;</span>
<span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.conf.Configuration;</span>
<span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.hbase.Cell;</span>
<span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.hbase.CellComparator;</span>
<span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.hbase.CellUtil;</span>
<span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.CompoundConfiguration;</span>
<span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.hbase.ExtendedCell;</span>
<span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.hbase.HConstants;</span>
<span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;</span>
<span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.hbase.client.TableDescriptor;</span>
<span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.io.hfile.HFile;</span>
<span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.hbase.util.ChecksumType;</span>
<span class="source-line-no">040</span><span id="line-40">import org.apache.yetus.audience.InterfaceAudience;</span>
<span class="source-line-no">041</span><span id="line-41">import org.slf4j.Logger;</span>
<span class="source-line-no">042</span><span id="line-42">import org.slf4j.LoggerFactory;</span>
<span class="source-line-no">043</span><span id="line-43"></span>
<span class="source-line-no">044</span><span id="line-44">/**</span>
<span class="source-line-no">045</span><span id="line-45"> * Utility functions for region server storage layer.</span>
<span class="source-line-no">046</span><span id="line-46"> */</span>
<span class="source-line-no">047</span><span id="line-47">@InterfaceAudience.Private</span>
<span class="source-line-no">048</span><span id="line-48">public final class StoreUtils {</span>
<span class="source-line-no">049</span><span id="line-49"></span>
<span class="source-line-no">050</span><span id="line-50"> private static final Logger LOG = LoggerFactory.getLogger(StoreUtils.class);</span>
<span class="source-line-no">051</span><span id="line-51"></span>
<span class="source-line-no">052</span><span id="line-52"> private StoreUtils() {</span>
<span class="source-line-no">053</span><span id="line-53"> }</span>
<span class="source-line-no">054</span><span id="line-54"></span>
<span class="source-line-no">055</span><span id="line-55"> /**</span>
<span class="source-line-no">056</span><span id="line-56"> * Creates a deterministic hash code for store file collection.</span>
<span class="source-line-no">057</span><span id="line-57"> */</span>
<span class="source-line-no">058</span><span id="line-58"> public static OptionalInt getDeterministicRandomSeed(Collection&lt;HStoreFile&gt; files) {</span>
<span class="source-line-no">059</span><span id="line-59"> return files.stream().mapToInt(f -&gt; f.getPath().getName().hashCode()).findFirst();</span>
<span class="source-line-no">060</span><span id="line-60"> }</span>
<span class="source-line-no">061</span><span id="line-61"></span>
<span class="source-line-no">062</span><span id="line-62"> /**</span>
<span class="source-line-no">063</span><span id="line-63"> * Determines whether any files in the collection are references.</span>
<span class="source-line-no">064</span><span id="line-64"> * @param files The files.</span>
<span class="source-line-no">065</span><span id="line-65"> */</span>
<span class="source-line-no">066</span><span id="line-66"> public static boolean hasReferences(Collection&lt;HStoreFile&gt; files) {</span>
<span class="source-line-no">067</span><span id="line-67"> // TODO: make sure that we won't pass null here in the future.</span>
<span class="source-line-no">068</span><span id="line-68"> return files != null &amp;&amp; files.stream().anyMatch(HStoreFile::isReference);</span>
<span class="source-line-no">069</span><span id="line-69"> }</span>
<span class="source-line-no">070</span><span id="line-70"></span>
<span class="source-line-no">071</span><span id="line-71"> /**</span>
<span class="source-line-no">072</span><span id="line-72"> * Gets lowest timestamp from candidate StoreFiles</span>
<span class="source-line-no">073</span><span id="line-73"> */</span>
<span class="source-line-no">074</span><span id="line-74"> public static long getLowestTimestamp(Collection&lt;HStoreFile&gt; candidates) throws IOException {</span>
<span class="source-line-no">075</span><span id="line-75"> long minTs = Long.MAX_VALUE;</span>
<span class="source-line-no">076</span><span id="line-76"> for (HStoreFile storeFile : candidates) {</span>
<span class="source-line-no">077</span><span id="line-77"> minTs = Math.min(minTs, storeFile.getModificationTimestamp());</span>
<span class="source-line-no">078</span><span id="line-78"> }</span>
<span class="source-line-no">079</span><span id="line-79"> return minTs;</span>
<span class="source-line-no">080</span><span id="line-80"> }</span>
<span class="source-line-no">081</span><span id="line-81"></span>
<span class="source-line-no">082</span><span id="line-82"> /**</span>
<span class="source-line-no">083</span><span id="line-83"> * Gets the largest file (with reader) out of the list of files.</span>
<span class="source-line-no">084</span><span id="line-84"> * @param candidates The files to choose from.</span>
<span class="source-line-no">085</span><span id="line-85"> * @return The largest file; null if no file has a reader.</span>
<span class="source-line-no">086</span><span id="line-86"> */</span>
<span class="source-line-no">087</span><span id="line-87"> static Optional&lt;HStoreFile&gt; getLargestFile(Collection&lt;HStoreFile&gt; candidates) {</span>
<span class="source-line-no">088</span><span id="line-88"> return candidates.stream().filter(f -&gt; f.getReader() != null)</span>
<span class="source-line-no">089</span><span id="line-89"> .max((f1, f2) -&gt; Long.compare(f1.getReader().length(), f2.getReader().length()));</span>
<span class="source-line-no">090</span><span id="line-90"> }</span>
<span class="source-line-no">091</span><span id="line-91"></span>
<span class="source-line-no">092</span><span id="line-92"> /**</span>
<span class="source-line-no">093</span><span id="line-93"> * Return the largest memstoreTS found across all storefiles in the given list. Store files that</span>
<span class="source-line-no">094</span><span id="line-94"> * were created by a mapreduce bulk load are ignored, as they do not correspond to any specific</span>
<span class="source-line-no">095</span><span id="line-95"> * put operation, and thus do not have a memstoreTS associated with them.</span>
<span class="source-line-no">096</span><span id="line-96"> */</span>
<span class="source-line-no">097</span><span id="line-97"> public static OptionalLong getMaxMemStoreTSInList(Collection&lt;HStoreFile&gt; sfs) {</span>
<span class="source-line-no">098</span><span id="line-98"> return sfs.stream().filter(sf -&gt; !sf.isBulkLoadResult()).mapToLong(HStoreFile::getMaxMemStoreTS)</span>
<span class="source-line-no">099</span><span id="line-99"> .max();</span>
<span class="source-line-no">100</span><span id="line-100"> }</span>
<span class="source-line-no">101</span><span id="line-101"></span>
<span class="source-line-no">102</span><span id="line-102"> /**</span>
<span class="source-line-no">103</span><span id="line-103"> * Return the highest sequence ID found across all storefiles in the given list.</span>
<span class="source-line-no">104</span><span id="line-104"> */</span>
<span class="source-line-no">105</span><span id="line-105"> public static OptionalLong getMaxSequenceIdInList(Collection&lt;HStoreFile&gt; sfs) {</span>
<span class="source-line-no">106</span><span id="line-106"> return sfs.stream().mapToLong(HStoreFile::getMaxSequenceId).max();</span>
<span class="source-line-no">107</span><span id="line-107"> }</span>
<span class="source-line-no">108</span><span id="line-108"></span>
<span class="source-line-no">109</span><span id="line-109"> /**</span>
<span class="source-line-no">110</span><span id="line-110"> * Gets the approximate mid-point of the given file that is optimal for use in splitting it.</span>
<span class="source-line-no">111</span><span id="line-111"> * @param file the store file</span>
<span class="source-line-no">112</span><span id="line-112"> * @param comparator Comparator used to compare KVs.</span>
<span class="source-line-no">113</span><span id="line-113"> * @return The split point row, or null if splitting is not possible, or reader is null.</span>
<span class="source-line-no">114</span><span id="line-114"> */</span>
<span class="source-line-no">115</span><span id="line-115"> static Optional&lt;byte[]&gt; getFileSplitPoint(HStoreFile file, CellComparator comparator)</span>
<span class="source-line-no">116</span><span id="line-116"> throws IOException {</span>
<span class="source-line-no">117</span><span id="line-117"> StoreFileReader reader = file.getReader();</span>
<span class="source-line-no">118</span><span id="line-118"> if (reader == null) {</span>
<span class="source-line-no">119</span><span id="line-119"> LOG.warn("Storefile " + file + " Reader is null; cannot get split point");</span>
<span class="source-line-no">120</span><span id="line-120"> return Optional.empty();</span>
<span class="source-line-no">121</span><span id="line-121"> }</span>
<span class="source-line-no">122</span><span id="line-122"> // Get first, last, and mid keys. Midkey is the key that starts block</span>
<span class="source-line-no">123</span><span id="line-123"> // in middle of hfile. Has column and timestamp. Need to return just</span>
<span class="source-line-no">124</span><span id="line-124"> // the row we want to split on as midkey.</span>
<span class="source-line-no">125</span><span id="line-125"> Optional&lt;ExtendedCell&gt; optionalMidKey = reader.midKey();</span>
<span class="source-line-no">126</span><span id="line-126"> if (!optionalMidKey.isPresent()) {</span>
<span class="source-line-no">127</span><span id="line-127"> return Optional.empty();</span>
<span class="source-line-no">128</span><span id="line-128"> }</span>
<span class="source-line-no">129</span><span id="line-129"> Cell midKey = optionalMidKey.get();</span>
<span class="source-line-no">130</span><span id="line-130"> Cell firstKey = reader.getFirstKey().get();</span>
<span class="source-line-no">131</span><span id="line-131"> Cell lastKey = reader.getLastKey().get();</span>
<span class="source-line-no">132</span><span id="line-132"> // if the midkey is the same as the first or last keys, we cannot (ever) split this region.</span>
<span class="source-line-no">133</span><span id="line-133"> if (</span>
<span class="source-line-no">134</span><span id="line-134"> comparator.compareRows(midKey, firstKey) == 0 || comparator.compareRows(midKey, lastKey) == 0</span>
<span class="source-line-no">135</span><span id="line-135"> ) {</span>
<span class="source-line-no">136</span><span id="line-136"> if (LOG.isDebugEnabled()) {</span>
<span class="source-line-no">137</span><span id="line-137"> LOG.debug("cannot split {} because midkey is the same as first or last row", file);</span>
<span class="source-line-no">138</span><span id="line-138"> }</span>
<span class="source-line-no">139</span><span id="line-139"> return Optional.empty();</span>
<span class="source-line-no">140</span><span id="line-140"> }</span>
<span class="source-line-no">141</span><span id="line-141"> return Optional.of(CellUtil.cloneRow(midKey));</span>
<span class="source-line-no">142</span><span id="line-142"> }</span>
<span class="source-line-no">143</span><span id="line-143"></span>
<span class="source-line-no">144</span><span id="line-144"> /**</span>
<span class="source-line-no">145</span><span id="line-145"> * Gets the mid point of the largest file passed in as split point.</span>
<span class="source-line-no">146</span><span id="line-146"> */</span>
<span class="source-line-no">147</span><span id="line-147"> static Optional&lt;byte[]&gt; getSplitPoint(Collection&lt;HStoreFile&gt; storefiles,</span>
<span class="source-line-no">148</span><span id="line-148"> CellComparator comparator) throws IOException {</span>
<span class="source-line-no">149</span><span id="line-149"> Optional&lt;HStoreFile&gt; largestFile = StoreUtils.getLargestFile(storefiles);</span>
<span class="source-line-no">150</span><span id="line-150"> return largestFile.isPresent()</span>
<span class="source-line-no">151</span><span id="line-151"> ? StoreUtils.getFileSplitPoint(largestFile.get(), comparator)</span>
<span class="source-line-no">152</span><span id="line-152"> : Optional.empty();</span>
<span class="source-line-no">153</span><span id="line-153"> }</span>
<span class="source-line-no">154</span><span id="line-154"></span>
<span class="source-line-no">155</span><span id="line-155"> /**</span>
<span class="source-line-no">156</span><span id="line-156"> * Returns the configured checksum algorithm.</span>
<span class="source-line-no">157</span><span id="line-157"> * @param conf The configuration</span>
<span class="source-line-no">158</span><span id="line-158"> * @return The checksum algorithm that is set in the configuration</span>
<span class="source-line-no">159</span><span id="line-159"> */</span>
<span class="source-line-no">160</span><span id="line-160"> public static ChecksumType getChecksumType(Configuration conf) {</span>
<span class="source-line-no">161</span><span id="line-161"> return ChecksumType.nameToType(</span>
<span class="source-line-no">162</span><span id="line-162"> conf.get(HConstants.CHECKSUM_TYPE_NAME, ChecksumType.getDefaultChecksumType().getName()));</span>
<span class="source-line-no">163</span><span id="line-163"> }</span>
<span class="source-line-no">164</span><span id="line-164"></span>
<span class="source-line-no">165</span><span id="line-165"> /**</span>
<span class="source-line-no">166</span><span id="line-166"> * Returns the configured bytesPerChecksum value.</span>
<span class="source-line-no">167</span><span id="line-167"> * @param conf The configuration</span>
<span class="source-line-no">168</span><span id="line-168"> * @return The bytesPerChecksum that is set in the configuration</span>
<span class="source-line-no">169</span><span id="line-169"> */</span>
<span class="source-line-no">170</span><span id="line-170"> public static int getBytesPerChecksum(Configuration conf) {</span>
<span class="source-line-no">171</span><span id="line-171"> return conf.getInt(HConstants.BYTES_PER_CHECKSUM, HFile.DEFAULT_BYTES_PER_CHECKSUM);</span>
<span class="source-line-no">172</span><span id="line-172"> }</span>
<span class="source-line-no">173</span><span id="line-173"></span>
<span class="source-line-no">174</span><span id="line-174"> public static Configuration createStoreConfiguration(Configuration conf, TableDescriptor td,</span>
<span class="source-line-no">175</span><span id="line-175"> ColumnFamilyDescriptor cfd) {</span>
<span class="source-line-no">176</span><span id="line-176"> // CompoundConfiguration will look for keys in reverse order of addition, so we'd</span>
<span class="source-line-no">177</span><span id="line-177"> // add global config first, then table and cf overrides, then cf metadata.</span>
<span class="source-line-no">178</span><span id="line-178"> return new CompoundConfiguration().add(conf).addBytesMap(td.getValues())</span>
<span class="source-line-no">179</span><span id="line-179"> .addStringMap(cfd.getConfiguration()).addBytesMap(cfd.getValues());</span>
<span class="source-line-no">180</span><span id="line-180"> }</span>
<span class="source-line-no">181</span><span id="line-181"></span>
<span class="source-line-no">182</span><span id="line-182"> public static List&lt;StoreFileInfo&gt; toStoreFileInfo(Collection&lt;HStoreFile&gt; storefiles) {</span>
<span class="source-line-no">183</span><span id="line-183"> return storefiles.stream().map(HStoreFile::getFileInfo).collect(Collectors.toList());</span>
<span class="source-line-no">184</span><span id="line-184"> }</span>
<span class="source-line-no">185</span><span id="line-185"></span>
<span class="source-line-no">186</span><span id="line-186"> public static long getTotalUncompressedBytes(List&lt;HStoreFile&gt; files) {</span>
<span class="source-line-no">187</span><span id="line-187"> return files.stream()</span>
<span class="source-line-no">188</span><span id="line-188"> .mapToLong(file -&gt; getStorefileFieldSize(file, StoreFileReader::getTotalUncompressedBytes))</span>
<span class="source-line-no">189</span><span id="line-189"> .sum();</span>
<span class="source-line-no">190</span><span id="line-190"> }</span>
<span class="source-line-no">191</span><span id="line-191"></span>
<span class="source-line-no">192</span><span id="line-192"> public static long getStorefilesSize(Collection&lt;HStoreFile&gt; files,</span>
<span class="source-line-no">193</span><span id="line-193"> Predicate&lt;HStoreFile&gt; predicate) {</span>
<span class="source-line-no">194</span><span id="line-194"> return files.stream().filter(predicate)</span>
<span class="source-line-no">195</span><span id="line-195"> .mapToLong(file -&gt; getStorefileFieldSize(file, StoreFileReader::length)).sum();</span>
<span class="source-line-no">196</span><span id="line-196"> }</span>
<span class="source-line-no">197</span><span id="line-197"></span>
<span class="source-line-no">198</span><span id="line-198"> public static long getStorefileFieldSize(HStoreFile file, ToLongFunction&lt;StoreFileReader&gt; f) {</span>
<span class="source-line-no">199</span><span id="line-199"> if (file == null) {</span>
<span class="source-line-no">200</span><span id="line-200"> return 0L;</span>
<span class="source-line-no">201</span><span id="line-201"> }</span>
<span class="source-line-no">202</span><span id="line-202"> StoreFileReader reader = file.getReader();</span>
<span class="source-line-no">203</span><span id="line-203"> if (reader == null) {</span>
<span class="source-line-no">204</span><span id="line-204"> return 0L;</span>
<span class="source-line-no">205</span><span id="line-205"> }</span>
<span class="source-line-no">206</span><span id="line-206"> return f.applyAsLong(reader);</span>
<span class="source-line-no">207</span><span id="line-207"> }</span>
<span class="source-line-no">208</span><span id="line-208">}</span>
</pre>
</div>
</main>
</body>
</html>