http://git-wip-us.apache.org/repos/asf/hbase-site/blob/1241ee85/devapidocs/src-html/org/apache/hadoop/hbase/regionserver/StoreScanner.StoreScannerCompactionRace.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/regionserver/StoreScanner.StoreScannerCompactionRace.html b/devapidocs/src-html/org/apache/hadoop/hbase/regionserver/StoreScanner.StoreScannerCompactionRace.html index 573de43..4e9c7c3 100644 --- a/devapidocs/src-html/org/apache/hadoop/hbase/regionserver/StoreScanner.StoreScannerCompactionRace.html +++ b/devapidocs/src-html/org/apache/hadoop/hbase/regionserver/StoreScanner.StoreScannerCompactionRace.html @@ -32,1012 +32,1081 @@ <span class="sourceLineNo">024</span>import java.io.IOException;<a name="line.24"></a> <span class="sourceLineNo">025</span>import java.io.InterruptedIOException;<a name="line.25"></a> <span class="sourceLineNo">026</span>import java.util.ArrayList;<a name="line.26"></a> -<span class="sourceLineNo">027</span>import java.util.List;<a name="line.27"></a> -<span class="sourceLineNo">028</span>import java.util.NavigableSet;<a name="line.28"></a> -<span class="sourceLineNo">029</span>import java.util.concurrent.CountDownLatch;<a name="line.29"></a> -<span class="sourceLineNo">030</span>import java.util.concurrent.locks.ReentrantLock;<a name="line.30"></a> -<span class="sourceLineNo">031</span><a name="line.31"></a> -<span class="sourceLineNo">032</span>import org.apache.commons.logging.Log;<a name="line.32"></a> -<span class="sourceLineNo">033</span>import org.apache.commons.logging.LogFactory;<a name="line.33"></a> -<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.Cell;<a name="line.34"></a> -<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.35"></a> -<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.36"></a> -<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.DoNotRetryIOException;<a name="line.37"></a> -<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.HConstants;<a name="line.38"></a> -<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.39"></a> -<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.40"></a> -<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.41"></a> -<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.client.IsolationLevel;<a name="line.42"></a> -<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.43"></a> -<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.executor.ExecutorService;<a name="line.44"></a> -<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.filter.Filter;<a name="line.45"></a> -<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.regionserver.ScannerContext.LimitScope;<a name="line.46"></a> -<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.regionserver.ScannerContext.NextState;<a name="line.47"></a> -<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.regionserver.handler.ParallelSeekHandler;<a name="line.48"></a> -<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.regionserver.querymatcher.CompactionScanQueryMatcher;<a name="line.49"></a> -<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.regionserver.querymatcher.LegacyScanQueryMatcher;<a name="line.50"></a> -<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.regionserver.querymatcher.ScanQueryMatcher;<a name="line.51"></a> -<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.regionserver.querymatcher.UserScanQueryMatcher;<a name="line.52"></a> -<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.53"></a> -<span class="sourceLineNo">054</span><a name="line.54"></a> -<span class="sourceLineNo">055</span>/**<a name="line.55"></a> -<span class="sourceLineNo">056</span> * Scanner scans both the memstore and the Store. Coalesce KeyValue stream<a name="line.56"></a> -<span class="sourceLineNo">057</span> * into List&lt;KeyValue&gt; for a single row.<a name="line.57"></a> -<span class="sourceLineNo">058</span> */<a name="line.58"></a> -<span class="sourceLineNo">059</span>@InterfaceAudience.Private<a name="line.59"></a> -<span class="sourceLineNo">060</span>public class StoreScanner extends NonReversedNonLazyKeyValueScanner<a name="line.60"></a> -<span class="sourceLineNo">061</span> implements KeyValueScanner, InternalScanner, ChangedReadersObserver {<a name="line.61"></a> -<span class="sourceLineNo">062</span> private static final Log LOG = LogFactory.getLog(StoreScanner.class);<a name="line.62"></a> -<span class="sourceLineNo">063</span> // In unit tests, the store could be null<a name="line.63"></a> -<span class="sourceLineNo">064</span> protected final Store store;<a name="line.64"></a> -<span class="sourceLineNo">065</span> protected ScanQueryMatcher matcher;<a name="line.65"></a> -<span class="sourceLineNo">066</span> protected KeyValueHeap heap;<a name="line.66"></a> -<span class="sourceLineNo">067</span> protected boolean cacheBlocks;<a name="line.67"></a> -<span class="sourceLineNo">068</span><a name="line.68"></a> -<span class="sourceLineNo">069</span> protected long countPerRow = 0;<a name="line.69"></a> -<span class="sourceLineNo">070</span> protected int storeLimit = -1;<a name="line.70"></a> -<span class="sourceLineNo">071</span> protected int storeOffset = 0;<a name="line.71"></a> -<span class="sourceLineNo">072</span><a name="line.72"></a> -<span class="sourceLineNo">073</span> // Used to indicate that the scanner has closed (see HBASE-1107)<a name="line.73"></a> -<span class="sourceLineNo">074</span> // Doesnt need to be volatile because it's always accessed via synchronized methods<a name="line.74"></a> -<span class="sourceLineNo">075</span> protected boolean closing = false;<a name="line.75"></a> -<span class="sourceLineNo">076</span> protected final boolean get;<a name="line.76"></a> -<span class="sourceLineNo">077</span> protected final boolean explicitColumnQuery;<a name="line.77"></a> -<span class="sourceLineNo">078</span> protected final boolean useRowColBloom;<a name="line.78"></a> -<span class="sourceLineNo">079</span> /**<a name="line.79"></a> -<span class="sourceLineNo">080</span> * A flag that enables StoreFileScanner parallel-seeking<a name="line.80"></a> -<span class="sourceLineNo">081</span> */<a name="line.81"></a> -<span class="sourceLineNo">082</span> protected boolean parallelSeekEnabled = false;<a name="line.82"></a> -<span class="sourceLineNo">083</span> protected ExecutorService executor;<a name="line.83"></a> -<span class="sourceLineNo">084</span> protected final Scan scan;<a name="line.84"></a> -<span class="sourceLineNo">085</span> protected final NavigableSet<byte[]> columns;<a name="line.85"></a> -<span class="sourceLineNo">086</span> protected final long oldestUnexpiredTS;<a name="line.86"></a> -<span class="sourceLineNo">087</span> protected final long now;<a name="line.87"></a> -<span class="sourceLineNo">088</span> protected final int minVersions;<a name="line.88"></a> -<span class="sourceLineNo">089</span> protected final long maxRowSize;<a name="line.89"></a> -<span class="sourceLineNo">090</span> protected final long cellsPerHeartbeatCheck;<a name="line.90"></a> -<span class="sourceLineNo">091</span><a name="line.91"></a> -<span class="sourceLineNo">092</span> // Collects all the KVHeap that are eagerly getting closed during the<a name="line.92"></a> -<span class="sourceLineNo">093</span> // course of a scan<a name="line.93"></a> -<span class="sourceLineNo">094</span> protected List<KeyValueHeap> heapsForDelayedClose = new ArrayList<>();<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> * The number of KVs seen by the scanner. Includes explicitly skipped KVs, but not<a name="line.97"></a> -<span class="sourceLineNo">098</span> * KVs skipped via seeking to next row/column. TODO: estimate them?<a name="line.98"></a> -<span class="sourceLineNo">099</span> */<a name="line.99"></a> -<span class="sourceLineNo">100</span> private long kvsScanned = 0;<a name="line.100"></a> -<span class="sourceLineNo">101</span> private Cell prevCell = null;<a name="line.101"></a> -<span class="sourceLineNo">102</span><a name="line.102"></a> -<span class="sourceLineNo">103</span> /** We don't ever expect to change this, the constant is just for clarity. */<a name="line.103"></a> -<span class="sourceLineNo">104</span> static final boolean LAZY_SEEK_ENABLED_BY_DEFAULT = true;<a name="line.104"></a> -<span class="sourceLineNo">105</span> public static final String STORESCANNER_PARALLEL_SEEK_ENABLE =<a name="line.105"></a> -<span class="sourceLineNo">106</span> "hbase.storescanner.parallel.seek.enable";<a name="line.106"></a> +<span class="sourceLineNo">027</span>import java.util.HashMap;<a name="line.27"></a> +<span class="sourceLineNo">028</span>import java.util.List;<a name="line.28"></a> +<span class="sourceLineNo">029</span>import java.util.Map;<a name="line.29"></a> +<span class="sourceLineNo">030</span>import java.util.NavigableSet;<a name="line.30"></a> +<span class="sourceLineNo">031</span>import java.util.concurrent.CountDownLatch;<a name="line.31"></a> +<span class="sourceLineNo">032</span>import java.util.concurrent.locks.ReentrantLock;<a name="line.32"></a> +<span class="sourceLineNo">033</span><a name="line.33"></a> +<span class="sourceLineNo">034</span>import org.apache.commons.logging.Log;<a name="line.34"></a> +<span class="sourceLineNo">035</span>import org.apache.commons.logging.LogFactory;<a name="line.35"></a> +<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.Cell;<a name="line.36"></a> +<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.37"></a> +<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.38"></a> +<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.DoNotRetryIOException;<a name="line.39"></a> +<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.HConstants;<a name="line.40"></a> +<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.41"></a> +<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.42"></a> +<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.43"></a> +<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.client.IsolationLevel;<a name="line.44"></a> +<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.45"></a> +<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.executor.ExecutorService;<a name="line.46"></a> +<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.filter.Filter;<a name="line.47"></a> +<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.regionserver.ScannerContext.LimitScope;<a name="line.48"></a> +<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.regionserver.ScannerContext.NextState;<a name="line.49"></a> +<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.regionserver.handler.ParallelSeekHandler;<a name="line.50"></a> +<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.regionserver.querymatcher.CompactionScanQueryMatcher;<a name="line.51"></a> +<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.regionserver.querymatcher.LegacyScanQueryMatcher;<a name="line.52"></a> +<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.regionserver.querymatcher.ScanQueryMatcher;<a name="line.53"></a> +<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.regionserver.querymatcher.UserScanQueryMatcher;<a name="line.54"></a> +<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.55"></a> +<span class="sourceLineNo">056</span><a name="line.56"></a> +<span class="sourceLineNo">057</span>/**<a name="line.57"></a> +<span class="sourceLineNo">058</span> * Scanner scans both the memstore and the Store. Coalesce KeyValue stream into List&lt;KeyValue&gt;<a name="line.58"></a> +<span class="sourceLineNo">059</span> * for a single row.<a name="line.59"></a> +<span class="sourceLineNo">060</span> * <p><a name="line.60"></a> +<span class="sourceLineNo">061</span> * The implementation is not thread safe. So there will be no race between next and close. The only<a name="line.61"></a> +<span class="sourceLineNo">062</span> * exception is updateReaders, it will be called in the memstore flush thread to indicate that there<a name="line.62"></a> +<span class="sourceLineNo">063</span> * is a flush.<a name="line.63"></a> +<span class="sourceLineNo">064</span> */<a name="line.64"></a> +<span class="sourceLineNo">065</span>@InterfaceAudience.Private<a name="line.65"></a> +<span class="sourceLineNo">066</span>public class StoreScanner extends NonReversedNonLazyKeyValueScanner<a name="line.66"></a> +<span class="sourceLineNo">067</span> implements KeyValueScanner, InternalScanner, ChangedReadersObserver {<a name="line.67"></a> +<span class="sourceLineNo">068</span> private static final Log LOG = LogFactory.getLog(StoreScanner.class);<a name="line.68"></a> +<span class="sourceLineNo">069</span> // In unit tests, the store could be null<a name="line.69"></a> +<span class="sourceLineNo">070</span> protected final Store store;<a name="line.70"></a> +<span class="sourceLineNo">071</span> private ScanQueryMatcher matcher;<a name="line.71"></a> +<span class="sourceLineNo">072</span> protected KeyValueHeap heap;<a name="line.72"></a> +<span class="sourceLineNo">073</span> private boolean cacheBlocks;<a name="line.73"></a> +<span class="sourceLineNo">074</span><a name="line.74"></a> +<span class="sourceLineNo">075</span> private long countPerRow = 0;<a name="line.75"></a> +<span class="sourceLineNo">076</span> private int storeLimit = -1;<a name="line.76"></a> +<span class="sourceLineNo">077</span> private int storeOffset = 0;<a name="line.77"></a> +<span class="sourceLineNo">078</span><a name="line.78"></a> +<span class="sourceLineNo">079</span> // Used to indicate that the scanner has closed (see HBASE-1107)<a name="line.79"></a> +<span class="sourceLineNo">080</span> // Do not need to be volatile because it's always accessed via synchronized methods<a name="line.80"></a> +<span class="sourceLineNo">081</span> private boolean closing = false;<a name="line.81"></a> +<span class="sourceLineNo">082</span> private final boolean get;<a name="line.82"></a> +<span class="sourceLineNo">083</span> private final boolean explicitColumnQuery;<a name="line.83"></a> +<span class="sourceLineNo">084</span> private final boolean useRowColBloom;<a name="line.84"></a> +<span class="sourceLineNo">085</span> /**<a name="line.85"></a> +<span class="sourceLineNo">086</span> * A flag that enables StoreFileScanner parallel-seeking<a name="line.86"></a> +<span class="sourceLineNo">087</span> */<a name="line.87"></a> +<span class="sourceLineNo">088</span> private boolean parallelSeekEnabled = false;<a name="line.88"></a> +<span class="sourceLineNo">089</span> private ExecutorService executor;<a name="line.89"></a> +<span class="sourceLineNo">090</span> private final Scan scan;<a name="line.90"></a> +<span class="sourceLineNo">091</span> private final long oldestUnexpiredTS;<a name="line.91"></a> +<span class="sourceLineNo">092</span> private final long now;<a name="line.92"></a> +<span class="sourceLineNo">093</span> private final int minVersions;<a name="line.93"></a> +<span class="sourceLineNo">094</span> private final long maxRowSize;<a name="line.94"></a> +<span class="sourceLineNo">095</span> private final long cellsPerHeartbeatCheck;<a name="line.95"></a> +<span class="sourceLineNo">096</span><a name="line.96"></a> +<span class="sourceLineNo">097</span> // Collects all the KVHeap that are eagerly getting closed during the<a name="line.97"></a> +<span class="sourceLineNo">098</span> // course of a scan<a name="line.98"></a> +<span class="sourceLineNo">099</span> private final List<KeyValueHeap> heapsForDelayedClose = new ArrayList<>();<a name="line.99"></a> +<span class="sourceLineNo">100</span><a name="line.100"></a> +<span class="sourceLineNo">101</span> /**<a name="line.101"></a> +<span class="sourceLineNo">102</span> * The number of KVs seen by the scanner. Includes explicitly skipped KVs, but not<a name="line.102"></a> +<span class="sourceLineNo">103</span> * KVs skipped via seeking to next row/column. TODO: estimate them?<a name="line.103"></a> +<span class="sourceLineNo">104</span> */<a name="line.104"></a> +<span class="sourceLineNo">105</span> private long kvsScanned = 0;<a name="line.105"></a> +<span class="sourceLineNo">106</span> private Cell prevCell = null;<a name="line.106"></a> <span class="sourceLineNo">107</span><a name="line.107"></a> -<span class="sourceLineNo">108</span> /** Used during unit testing to ensure that lazy seek does save seek ops */<a name="line.108"></a> -<span class="sourceLineNo">109</span> protected static boolean lazySeekEnabledGlobally =<a name="line.109"></a> -<span class="sourceLineNo">110</span> LAZY_SEEK_ENABLED_BY_DEFAULT;<a name="line.110"></a> -<span class="sourceLineNo">111</span><a name="line.111"></a> -<span class="sourceLineNo">112</span> /**<a name="line.112"></a> -<span class="sourceLineNo">113</span> * The number of cells scanned in between timeout checks. Specifying a larger value means that<a name="line.113"></a> -<span class="sourceLineNo">114</span> * timeout checks will occur less frequently. Specifying a small value will lead to more frequent<a name="line.114"></a> -<span class="sourceLineNo">115</span> * timeout checks.<a name="line.115"></a> -<span class="sourceLineNo">116</span> */<a name="line.116"></a> -<span class="sourceLineNo">117</span> public static final String HBASE_CELLS_SCANNED_PER_HEARTBEAT_CHECK =<a name="line.117"></a> -<span class="sourceLineNo">118</span> "hbase.cells.scanned.per.heartbeat.check";<a name="line.118"></a> -<span class="sourceLineNo">119</span><a name="line.119"></a> -<span class="sourceLineNo">120</span> /**<a name="line.120"></a> -<span class="sourceLineNo">121</span> * Default value of {@link #HBASE_CELLS_SCANNED_PER_HEARTBEAT_CHECK}.<a name="line.121"></a> -<span class="sourceLineNo">122</span> */<a name="line.122"></a> -<span class="sourceLineNo">123</span> public static final long DEFAULT_HBASE_CELLS_SCANNED_PER_HEARTBEAT_CHECK = 10000;<a name="line.123"></a> -<span class="sourceLineNo">124</span><a name="line.124"></a> -<span class="sourceLineNo">125</span> // if heap == null and lastTop != null, you need to reseek given the key below<a name="line.125"></a> -<span class="sourceLineNo">126</span> protected Cell lastTop = null;<a name="line.126"></a> -<span class="sourceLineNo">127</span><a name="line.127"></a> -<span class="sourceLineNo">128</span> // A flag whether use pread for scan<a name="line.128"></a> -<span class="sourceLineNo">129</span> private final boolean scanUsePread;<a name="line.129"></a> -<span class="sourceLineNo">130</span> // Indicates whether there was flush during the course of the scan<a name="line.130"></a> -<span class="sourceLineNo">131</span> protected volatile boolean flushed = false;<a name="line.131"></a> -<span class="sourceLineNo">132</span> // generally we get one file from a flush<a name="line.132"></a> -<span class="sourceLineNo">133</span> protected List<StoreFile> flushedStoreFiles = new ArrayList<>(1);<a name="line.133"></a> -<span class="sourceLineNo">134</span> // The current list of scanners<a name="line.134"></a> -<span class="sourceLineNo">135</span> protected List<KeyValueScanner> currentScanners = new ArrayList<>();<a name="line.135"></a> -<span class="sourceLineNo">136</span> // flush update lock<a name="line.136"></a> -<span class="sourceLineNo">137</span> private ReentrantLock flushLock = new ReentrantLock();<a name="line.137"></a> +<span class="sourceLineNo">108</span> private final long preadMaxBytes;<a name="line.108"></a> +<span class="sourceLineNo">109</span> private long bytesRead;<a name="line.109"></a> +<span class="sourceLineNo">110</span><a name="line.110"></a> +<span class="sourceLineNo">111</span> /** We don't ever expect to change this, the constant is just for clarity. */<a name="line.111"></a> +<span class="sourceLineNo">112</span> static final boolean LAZY_SEEK_ENABLED_BY_DEFAULT = true;<a name="line.112"></a> +<span class="sourceLineNo">113</span> public static final String STORESCANNER_PARALLEL_SEEK_ENABLE =<a name="line.113"></a> +<span class="sourceLineNo">114</span> "hbase.storescanner.parallel.seek.enable";<a name="line.114"></a> +<span class="sourceLineNo">115</span><a name="line.115"></a> +<span class="sourceLineNo">116</span> /** Used during unit testing to ensure that lazy seek does save seek ops */<a name="line.116"></a> +<span class="sourceLineNo">117</span> private static boolean lazySeekEnabledGlobally = LAZY_SEEK_ENABLED_BY_DEFAULT;<a name="line.117"></a> +<span class="sourceLineNo">118</span><a name="line.118"></a> +<span class="sourceLineNo">119</span> /**<a name="line.119"></a> +<span class="sourceLineNo">120</span> * The number of cells scanned in between timeout checks. Specifying a larger value means that<a name="line.120"></a> +<span class="sourceLineNo">121</span> * timeout checks will occur less frequently. Specifying a small value will lead to more frequent<a name="line.121"></a> +<span class="sourceLineNo">122</span> * timeout checks.<a name="line.122"></a> +<span class="sourceLineNo">123</span> */<a name="line.123"></a> +<span class="sourceLineNo">124</span> public static final String HBASE_CELLS_SCANNED_PER_HEARTBEAT_CHECK =<a name="line.124"></a> +<span class="sourceLineNo">125</span> "hbase.cells.scanned.per.heartbeat.check";<a name="line.125"></a> +<span class="sourceLineNo">126</span><a name="line.126"></a> +<span class="sourceLineNo">127</span> /**<a name="line.127"></a> +<span class="sourceLineNo">128</span> * Default value of {@link #HBASE_CELLS_SCANNED_PER_HEARTBEAT_CHECK}.<a name="line.128"></a> +<span class="sourceLineNo">129</span> */<a name="line.129"></a> +<span class="sourceLineNo">130</span> public static final long DEFAULT_HBASE_CELLS_SCANNED_PER_HEARTBEAT_CHECK = 10000;<a name="line.130"></a> +<span class="sourceLineNo">131</span><a name="line.131"></a> +<span class="sourceLineNo">132</span> /**<a name="line.132"></a> +<span class="sourceLineNo">133</span> * If the read type if Scan.ReadType.DEFAULT, we will start with pread, and if the kvs we scanned<a name="line.133"></a> +<span class="sourceLineNo">134</span> * reaches this limit, we will reopen the scanner with stream. The default value is 4 times of<a name="line.134"></a> +<span class="sourceLineNo">135</span> * block size for this store.<a name="line.135"></a> +<span class="sourceLineNo">136</span> */<a name="line.136"></a> +<span class="sourceLineNo">137</span> public static final String STORESCANNER_PREAD_MAX_BYTES = "hbase.storescanner.pread.max.bytes";<a name="line.137"></a> <span class="sourceLineNo">138</span><a name="line.138"></a> -<span class="sourceLineNo">139</span> protected final long readPt;<a name="line.139"></a> +<span class="sourceLineNo">139</span> private final Scan.ReadType readType;<a name="line.139"></a> <span class="sourceLineNo">140</span><a name="line.140"></a> -<span class="sourceLineNo">141</span> // used by the injection framework to test race between StoreScanner construction and compaction<a name="line.141"></a> -<span class="sourceLineNo">142</span> enum StoreScannerCompactionRace {<a name="line.142"></a> -<span class="sourceLineNo">143</span> BEFORE_SEEK,<a name="line.143"></a> -<span class="sourceLineNo">144</span> AFTER_SEEK,<a name="line.144"></a> -<span class="sourceLineNo">145</span> COMPACT_COMPLETE<a name="line.145"></a> -<span class="sourceLineNo">146</span> }<a name="line.146"></a> -<span class="sourceLineNo">147</span><a name="line.147"></a> -<span class="sourceLineNo">148</span> /** An internal constructor. */<a name="line.148"></a> -<span class="sourceLineNo">149</span> protected StoreScanner(Store store, Scan scan, final ScanInfo scanInfo,<a name="line.149"></a> -<span class="sourceLineNo">150</span> final NavigableSet<byte[]> columns, long readPt, boolean cacheBlocks) {<a name="line.150"></a> -<span class="sourceLineNo">151</span> this.readPt = readPt;<a name="line.151"></a> -<span class="sourceLineNo">152</span> this.store = store;<a name="line.152"></a> -<span class="sourceLineNo">153</span> this.cacheBlocks = cacheBlocks;<a name="line.153"></a> -<span class="sourceLineNo">154</span> get = scan.isGetScan();<a name="line.154"></a> -<span class="sourceLineNo">155</span> int numCol = columns == null ? 0 : columns.size();<a name="line.155"></a> -<span class="sourceLineNo">156</span> explicitColumnQuery = numCol > 0;<a name="line.156"></a> -<span class="sourceLineNo">157</span> this.scan = scan;<a name="line.157"></a> -<span class="sourceLineNo">158</span> this.columns = columns;<a name="line.158"></a> -<span class="sourceLineNo">159</span> this.now = EnvironmentEdgeManager.currentTime();<a name="line.159"></a> -<span class="sourceLineNo">160</span> this.oldestUnexpiredTS = scan.isRaw() ? 0L : now - scanInfo.getTtl();<a name="line.160"></a> -<span class="sourceLineNo">161</span> this.minVersions = scanInfo.getMinVersions();<a name="line.161"></a> -<span class="sourceLineNo">162</span><a name="line.162"></a> -<span class="sourceLineNo">163</span> // We look up row-column Bloom filters for multi-column queries as part of<a name="line.163"></a> -<span class="sourceLineNo">164</span> // the seek operation. However, we also look the row-column Bloom filter<a name="line.164"></a> -<span class="sourceLineNo">165</span> // for multi-row (non-"get") scans because this is not done in<a name="line.165"></a> -<span class="sourceLineNo">166</span> // StoreFile.passesBloomFilter(Scan, SortedSet<byte[]>).<a name="line.166"></a> -<span class="sourceLineNo">167</span> this.useRowColBloom = numCol > 1 || (!get && numCol == 1);<a name="line.167"></a> -<span class="sourceLineNo">168</span><a name="line.168"></a> -<span class="sourceLineNo">169</span> this.maxRowSize = scanInfo.getTableMaxRowSize();<a name="line.169"></a> -<span class="sourceLineNo">170</span> if (get) {<a name="line.170"></a> -<span class="sourceLineNo">171</span> this.scanUsePread = true;<a name="line.171"></a> -<span class="sourceLineNo">172</span> } else {<a name="line.172"></a> -<span class="sourceLineNo">173</span> switch (scan.getReadType()) {<a name="line.173"></a> -<span class="sourceLineNo">174</span> case STREAM:<a name="line.174"></a> -<span class="sourceLineNo">175</span> this.scanUsePread = false;<a name="line.175"></a> -<span class="sourceLineNo">176</span> break;<a name="line.176"></a> -<span class="sourceLineNo">177</span> case PREAD:<a name="line.177"></a> -<span class="sourceLineNo">178</span> this.scanUsePread = true;<a name="line.178"></a> -<span class="sourceLineNo">179</span> break;<a name="line.179"></a> -<span class="sourceLineNo">180</span> default:<a name="line.180"></a> -<span class="sourceLineNo">181</span> this.scanUsePread = scanInfo.isUsePread();<a name="line.181"></a> -<span class="sourceLineNo">182</span> break;<a name="line.182"></a> -<span class="sourceLineNo">183</span> }<a name="line.183"></a> -<span class="sourceLineNo">184</span> }<a name="line.184"></a> -<span class="sourceLineNo">185</span> this.cellsPerHeartbeatCheck = scanInfo.getCellsPerTimeoutCheck();<a name="line.185"></a> -<span class="sourceLineNo">186</span> // Parallel seeking is on if the config allows and more there is more than one store file.<a name="line.186"></a> -<span class="sourceLineNo">187</span> if (this.store != null && this.store.getStorefilesCount() > 1) {<a name="line.187"></a> -<span class="sourceLineNo">188</span> RegionServerServices rsService = ((HStore)store).getHRegion().getRegionServerServices();<a name="line.188"></a> -<span class="sourceLineNo">189</span> if (rsService != null && scanInfo.isParallelSeekEnabled()) {<a name="line.189"></a> -<span class="sourceLineNo">190</span> this.parallelSeekEnabled = true;<a name="line.190"></a> -<span class="sourceLineNo">191</span> this.executor = rsService.getExecutorService();<a name="line.191"></a> -<span class="sourceLineNo">192</span> }<a name="line.192"></a> -<span class="sourceLineNo">193</span> }<a name="line.193"></a> -<span class="sourceLineNo">194</span> }<a name="line.194"></a> -<span class="sourceLineNo">195</span><a name="line.195"></a> -<span class="sourceLineNo">196</span> protected void addCurrentScanners(List<? extends KeyValueScanner> scanners) {<a name="line.196"></a> -<span class="sourceLineNo">197</span> this.currentScanners.addAll(scanners);<a name="line.197"></a> -<span class="sourceLineNo">198</span> }<a name="line.198"></a> -<span class="sourceLineNo">199</span><a name="line.199"></a> -<span class="sourceLineNo">200</span> /**<a name="line.200"></a> -<span class="sourceLineNo">201</span> * Opens a scanner across memstore, snapshot, and all StoreFiles. Assumes we<a name="line.201"></a> -<span class="sourceLineNo">202</span> * are not in a compaction.<a name="line.202"></a> -<span class="sourceLineNo">203</span> *<a name="line.203"></a> -<span class="sourceLineNo">204</span> * @param store who we scan<a name="line.204"></a> -<span class="sourceLineNo">205</span> * @param scan the spec<a name="line.205"></a> -<span class="sourceLineNo">206</span> * @param columns which columns we are scanning<a name="line.206"></a> -<span class="sourceLineNo">207</span> * @throws IOException<a name="line.207"></a> -<span class="sourceLineNo">208</span> */<a name="line.208"></a> -<span class="sourceLineNo">209</span> public StoreScanner(Store store, ScanInfo scanInfo, Scan scan, final NavigableSet<byte[]> columns,<a name="line.209"></a> -<span class="sourceLineNo">210</span> long readPt)<a name="line.210"></a> -<span class="sourceLineNo">211</span> throws IOException {<a name="line.211"></a> -<span class="sourceLineNo">212</span> this(store, scan, scanInfo, columns, readPt, scan.getCacheBlocks());<a name="line.212"></a> -<span class="sourceLineNo">213</span> if (columns != null && scan.isRaw()) {<a name="line.213"></a> -<span class="sourceLineNo">214</span> throw new DoNotRetryIOException("Cannot specify any column for a raw scan");<a name="line.214"></a> -<span class="sourceLineNo">215</span> }<a name="line.215"></a> -<span class="sourceLineNo">216</span> matcher = UserScanQueryMatcher.create(scan, scanInfo, columns, oldestUnexpiredTS, now,<a name="line.216"></a> -<span class="sourceLineNo">217</span> store.getCoprocessorHost());<a name="line.217"></a> -<span class="sourceLineNo">218</span><a name="line.218"></a> -<span class="sourceLineNo">219</span> this.store.addChangedReaderObserver(this);<a name="line.219"></a> -<span class="sourceLineNo">220</span><a name="line.220"></a> -<span class="sourceLineNo">221</span> try {<a name="line.221"></a> -<span class="sourceLineNo">222</span> // Pass columns to try to filter out unnecessary StoreFiles.<a name="line.222"></a> -<span class="sourceLineNo">223</span> List<KeyValueScanner> scanners = getScannersNoCompaction();<a name="line.223"></a> -<span class="sourceLineNo">224</span><a name="line.224"></a> -<span class="sourceLineNo">225</span> // Seek all scanners to the start of the Row (or if the exact matching row<a name="line.225"></a> -<span class="sourceLineNo">226</span> // key does not exist, then to the start of the next matching Row).<a name="line.226"></a> -<span class="sourceLineNo">227</span> // Always check bloom filter to optimize the top row seek for delete<a name="line.227"></a> -<span class="sourceLineNo">228</span> // family marker.<a name="line.228"></a> -<span class="sourceLineNo">229</span> seekScanners(scanners, matcher.getStartKey(), explicitColumnQuery && lazySeekEnabledGlobally,<a name="line.229"></a> -<span class="sourceLineNo">230</span> parallelSeekEnabled);<a name="line.230"></a> -<span class="sourceLineNo">231</span><a name="line.231"></a> -<span class="sourceLineNo">232</span> // set storeLimit<a name="line.232"></a> -<span class="sourceLineNo">233</span> this.storeLimit = scan.getMaxResultsPerColumnFamily();<a name="line.233"></a> -<span class="sourceLineNo">234</span><a name="line.234"></a> -<span class="sourceLineNo">235</span> // set rowOffset<a name="line.235"></a> -<span class="sourceLineNo">236</span> this.storeOffset = scan.getRowOffsetPerColumnFamily();<a name="line.236"></a> -<span class="sourceLineNo">237</span> addCurrentScanners(scanners);<a name="line.237"></a> -<span class="sourceLineNo">238</span> // Combine all seeked scanners with a heap<a name="line.238"></a> -<span class="sourceLineNo">239</span> resetKVHeap(scanners, store.getComparator());<a name="line.239"></a> -<span class="sourceLineNo">240</span> } catch (IOException e) {<a name="line.240"></a> -<span class="sourceLineNo">241</span> // remove us from the HStore#changedReaderObservers here or we'll have no chance to<a name="line.241"></a> -<span class="sourceLineNo">242</span> // and might cause memory leak<a name="line.242"></a> -<span class="sourceLineNo">243</span> this.store.deleteChangedReaderObserver(this);<a name="line.243"></a> -<span class="sourceLineNo">244</span> throw e;<a name="line.244"></a> -<span class="sourceLineNo">245</span> }<a name="line.245"></a> -<span class="sourceLineNo">246</span> }<a name="line.246"></a> -<span class="sourceLineNo">247</span><a name="line.247"></a> -<span class="sourceLineNo">248</span> /**<a name="line.248"></a> -<span class="sourceLineNo">249</span> * Used for compactions.<p><a name="line.249"></a> -<span class="sourceLineNo">250</span> *<a name="line.250"></a> -<span class="sourceLineNo">251</span> * Opens a scanner across specified StoreFiles.<a name="line.251"></a> -<span class="sourceLineNo">252</span> * @param store who we scan<a name="line.252"></a> -<span class="sourceLineNo">253</span> * @param scan the spec<a name="line.253"></a> -<span class="sourceLineNo">254</span> * @param scanners ancillary scanners<a name="line.254"></a> -<span class="sourceLineNo">255</span> * @param smallestReadPoint the readPoint that we should use for tracking<a name="line.255"></a> -<span class="sourceLineNo">256</span> * versions<a name="line.256"></a> -<span class="sourceLineNo">257</span> */<a name="line.257"></a> -<span class="sourceLineNo">258</span> public StoreScanner(Store store, ScanInfo scanInfo, Scan scan,<a name="line.258"></a> -<span class="sourceLineNo">259</span> List<? extends KeyValueScanner> scanners, ScanType scanType,<a name="line.259"></a> -<span class="sourceLineNo">260</span> long smallestReadPoint, long earliestPutTs) throws IOException {<a name="line.260"></a> -<span class="sourceLineNo">261</span> this(store, scanInfo, scan, scanners, scanType, smallestReadPoint, earliestPutTs, null, null);<a name="line.261"></a> -<span class="sourceLineNo">262</span> }<a name="line.262"></a> -<span class="sourceLineNo">263</span><a name="line.263"></a> -<span class="sourceLineNo">264</span> /**<a name="line.264"></a> -<span class="sourceLineNo">265</span> * Used for compactions that drop deletes from a limited range of rows.<p><a name="line.265"></a> -<span class="sourceLineNo">266</span> *<a name="line.266"></a> -<span class="sourceLineNo">267</span> * Opens a scanner across specified StoreFiles.<a name="line.267"></a> -<span class="sourceLineNo">268</span> * @param store who we scan<a name="line.268"></a> -<span class="sourceLineNo">269</span> * @param scan the spec<a name="line.269"></a> -<span class="sourceLineNo">270</span> * @param scanners ancillary scanners<a name="line.270"></a> -<span class="sourceLineNo">271</span> * @param smallestReadPoint the readPoint that we should use for tracking versions<a name="line.271"></a> -<span class="sourceLineNo">272</span> * @param dropDeletesFromRow The inclusive left bound of the range; can be EMPTY_START_ROW.<a name="line.272"></a> -<span class="sourceLineNo">273</span> * @param dropDeletesToRow The exclusive right bound of the range; can be EMPTY_END_ROW.<a name="line.273"></a> -<span class="sourceLineNo">274</span> */<a name="line.274"></a> -<span class="sourceLineNo">275</span> public StoreScanner(Store store, ScanInfo scanInfo, Scan scan,<a name="line.275"></a> -<span class="sourceLineNo">276</span> List<? extends KeyValueScanner> scanners, long smallestReadPoint, long earliestPutTs,<a name="line.276"></a> -<span class="sourceLineNo">277</span> byte[] dropDeletesFromRow, byte[] dropDeletesToRow) throws IOException {<a name="line.277"></a> -<span class="sourceLineNo">278</span> this(store, scanInfo, scan, scanners, ScanType.COMPACT_RETAIN_DELETES, smallestReadPoint,<a name="line.278"></a> -<span class="sourceLineNo">279</span> earliestPutTs, dropDeletesFromRow, dropDeletesToRow);<a name="line.279"></a> -<span class="sourceLineNo">280</span> }<a name="line.280"></a> -<span class="sourceLineNo">281</span><a name="line.281"></a> -<span class="sourceLineNo">282</span> private StoreScanner(Store store, ScanInfo scanInfo, Scan scan,<a name="line.282"></a> -<span class="sourceLineNo">283</span> List<? extends KeyValueScanner> scanners, ScanType scanType, long smallestReadPoint,<a name="line.283"></a> -<span class="sourceLineNo">284</span> long earliestPutTs, byte[] dropDeletesFromRow, byte[] dropDeletesToRow) throws IOException {<a name="line.284"></a> -<span class="sourceLineNo">285</span> this(store, scan, scanInfo, null,<a name="line.285"></a> -<span class="sourceLineNo">286</span> ((HStore) store).getHRegion().getReadPoint(IsolationLevel.READ_COMMITTED), false);<a name="line.286"></a> -<span class="sourceLineNo">287</span> if (scan.hasFilter() || (scan.getStartRow() != null && scan.getStartRow().length > 0)<a name="line.287"></a> -<span class="sourceLineNo">288</span> || (scan.getStopRow() != null && scan.getStopRow().length > 0)<a name="line.288"></a> -<span class="sourceLineNo">289</span> || !scan.getTimeRange().isAllTime()) {<a name="line.289"></a> -<span class="sourceLineNo">290</span> // use legacy query matcher since we do not consider the scan object in our code. Only used to<a name="line.290"></a> -<span class="sourceLineNo">291</span> // keep compatibility for coprocessor.<a name="line.291"></a> -<span class="sourceLineNo">292</span> matcher = LegacyScanQueryMatcher.create(scan, scanInfo, null, scanType, smallestReadPoint,<a name="line.292"></a> -<span class="sourceLineNo">293</span> earliestPutTs, oldestUnexpiredTS, now, dropDeletesFromRow, dropDeletesToRow,<a name="line.293"></a> -<span class="sourceLineNo">294</span> store.getCoprocessorHost());<a name="line.294"></a> -<span class="sourceLineNo">295</span> } else {<a name="line.295"></a> -<span class="sourceLineNo">296</span> matcher = CompactionScanQueryMatcher.create(scanInfo, scanType, smallestReadPoint,<a name="line.296"></a> -<span class="sourceLineNo">297</span> earliestPutTs, oldestUnexpiredTS, now, dropDeletesFromRow, dropDeletesToRow,<a name="line.297"></a> -<span class="sourceLineNo">298</span> store.getCoprocessorHost());<a name="line.298"></a> -<span class="sourceLineNo">299</span> }<a name="line.299"></a> -<span class="sourceLineNo">300</span><a name="line.300"></a> -<span class="sourceLineNo">301</span> // Filter the list of scanners using Bloom filters, time range, TTL, etc.<a name="line.301"></a> -<span class="sourceLineNo">302</span> scanners = selectScannersFrom(scanners);<a name="line.302"></a> -<span class="sourceLineNo">303</span><a name="line.303"></a> -<span class="sourceLineNo">304</span> // Seek all scanners to the initial key<a name="line.304"></a> -<span class="sourceLineNo">305</span> seekScanners(scanners, matcher.getStartKey(), false, parallelSeekEnabled);<a name="line.305"></a> -<span class="sourceLineNo">306</span> addCurrentScanners(scanners);<a name="line.306"></a> -<span class="sourceLineNo">307</span> // Combine all seeked scanners with a heap<a name="line.307"></a> -<span class="sourceLineNo">308</span> resetKVHeap(scanners, store.getComparator());<a name="line.308"></a> -<span class="sourceLineNo">309</span> }<a name="line.309"></a> -<span class="sourceLineNo">310</span><a name="line.310"></a> -<span class="sourceLineNo">311</span> @VisibleForTesting<a name="line.311"></a> -<span class="sourceLineNo">312</span> StoreScanner(final Scan scan, ScanInfo scanInfo,<a name="line.312"></a> -<span class="sourceLineNo">313</span> ScanType scanType, final NavigableSet<byte[]> columns,<a name="line.313"></a> -<span class="sourceLineNo">314</span> final List<? extends KeyValueScanner> scanners) throws IOException {<a name="line.314"></a> -<span class="sourceLineNo">315</span> this(scan, scanInfo, scanType, columns, scanners,<a name="line.315"></a> -<span class="sourceLineNo">316</span> HConstants.LATEST_TIMESTAMP,<a name="line.316"></a> -<span class="sourceLineNo">317</span> // 0 is passed as readpoint because the test bypasses Store<a name="line.317"></a> -<span class="sourceLineNo">318</span> 0);<a name="line.318"></a> -<span class="sourceLineNo">319</span> }<a name="line.319"></a> -<span class="sourceLineNo">320</span><a name="line.320"></a> -<span class="sourceLineNo">321</span> @VisibleForTesting<a name="line.321"></a> -<span class="sourceLineNo">322</span> StoreScanner(final Scan scan, ScanInfo scanInfo,<a name="line.322"></a> -<span class="sourceLineNo">323</span> ScanType scanType, final NavigableSet<byte[]> columns,<a name="line.323"></a> -<span class="sourceLineNo">324</span> final List<? extends KeyValueScanner> scanners, long earliestPutTs)<a name="line.324"></a> -<span class="sourceLineNo">325</span> throws IOException {<a name="line.325"></a> -<span class="sourceLineNo">326</span> this(scan, scanInfo, scanType, columns, scanners, earliestPutTs,<a name="line.326"></a> -<span class="sourceLineNo">327</span> // 0 is passed as readpoint because the test bypasses Store<a name="line.327"></a> -<span class="sourceLineNo">328</span> 0);<a name="line.328"></a> -<span class="sourceLineNo">329</span> }<a name="line.329"></a> -<span class="sourceLineNo">330</span><a name="line.330"></a> -<span class="sourceLineNo">331</span> public StoreScanner(final Scan scan, ScanInfo scanInfo, ScanType scanType,<a name="line.331"></a> -<span class="sourceLineNo">332</span> final NavigableSet<byte[]> columns, final List<? extends KeyValueScanner> scanners, long earliestPutTs,<a name="line.332"></a> -<span class="sourceLineNo">333</span> long readPt) throws IOException {<a name="line.333"></a> -<span class="sourceLineNo">334</span> this(null, scan, scanInfo, columns, readPt,<a name="line.334"></a> -<span class="sourceLineNo">335</span> scanType == ScanType.USER_SCAN ? scan.getCacheBlocks() : false);<a name="line.335"></a> -<span class="sourceLineNo">336</span> if (scanType == ScanType.USER_SCAN) {<a name="line.336"></a> -<span class="sourceLineNo">337</span> this.matcher = UserScanQueryMatcher.create(scan, scanInfo, columns, oldestUnexpiredTS, now,<a name="line.337"></a> -<span class="sourceLineNo">338</span> null);<a name="line.338"></a> -<span class="sourceLineNo">339</span> } else {<a name="line.339"></a> -<span class="sourceLineNo">340</span> if (scan.hasFilter() || (scan.getStartRow() != null && scan.getStartRow().length > 0)<a name="line.340"></a> -<span class="sourceLineNo">341</span> || (scan.getStopRow() != null && scan.getStopRow().length > 0)<a name="line.341"></a> -<span class="sourceLineNo">342</span> || !scan.getTimeRange().isAllTime() || columns != null) {<a name="line.342"></a> -<span class="sourceLineNo">343</span> // use legacy query matcher since we do not consider the scan object in our code. Only used<a name="line.343"></a> -<span class="sourceLineNo">344</span> // to keep compatibility for coprocessor.<a name="line.344"></a> -<span class="sourceLineNo">345</span> matcher = LegacyScanQueryMatcher.create(scan, scanInfo, columns, scanType, Long.MAX_VALUE,<a name="line.345"></a> -<span class="sourceLineNo">346</span> earliestPutTs, oldestUnexpiredTS, now, null, null, store.getCoprocessorHost());<a name="line.346"></a> -<span class="sourceLineNo">347</span> } else {<a name="line.347"></a> -<span class="sourceLineNo">348</span> this.matcher = CompactionScanQueryMatcher.create(scanInfo, scanType, Long.MAX_VALUE,<a name="line.348"></a> -<span class="sourceLineNo">349</span> earliestPutTs, oldestUnexpiredTS, now, null, null, null);<a name="line.349"></a> -<span class="sourceLineNo">350</span> }<a name="line.350"></a> -<span class="sourceLineNo">351</span> }<a name="line.351"></a> -<span class="sourceLineNo">352</span><a name="line.352"></a> -<span class="sourceLineNo">353</span> // Seek all scanners to the initial key<a name="line.353"></a> -<span class="sourceLineNo">354</span> seekScanners(scanners, matcher.getStartKey(), false, parallelSeekEnabled);<a name="line.354"></a> -<span class="sourceLineNo">355</span> addCurrentScanners(scanners);<a name="line.355"></a> -<span class="sourceLineNo">356</span> resetKVHeap(scanners, scanInfo.getComparator());<a name="line.356"></a> -<span class="sourceLineNo">357</span> }<a name="line.357"></a> -<span class="sourceLineNo">358</span><a name="line.358"></a> -<span class="sourceLineNo">359</span> /**<a name="line.359"></a> -<span class="sourceLineNo">360</span> * Get a filtered list of scanners. Assumes we are not in a compaction.<a name="line.360"></a> -<span class="sourceLineNo">361</span> * @return list of scanners to seek<a name="line.361"></a> -<span class="sourceLineNo">362</span> */<a name="line.362"></a> -<span class="sourceLineNo">363</span> protected List<KeyValueScanner> getScannersNoCompaction() throws IOException {<a name="line.363"></a> -<span class="sourceLineNo">364</span> return selectScannersFrom(store.getScanners(cacheBlocks, get, scanUsePread, false, matcher,<a name="line.364"></a> -<span class="sourceLineNo">365</span> scan.getStartRow(), scan.getStopRow(), this.readPt));<a name="line.365"></a> -<span class="sourceLineNo">366</span> }<a name="line.366"></a> -<span class="sourceLineNo">367</span><a name="line.367"></a> -<span class="sourceLineNo">368</span> /**<a name="line.368"></a> -<span class="sourceLineNo">369</span> * Seek the specified scanners with the given key<a name="line.369"></a> -<span class="sourceLineNo">370</span> * @param scanners<a name="line.370"></a> -<span class="sourceLineNo">371</span> * @param seekKey<a name="line.371"></a> -<span class="sourceLineNo">372</span> * @param isLazy true if using lazy seek<a name="line.372"></a> -<span class="sourceLineNo">373</span> * @param isParallelSeek true if using parallel seek<a name="line.373"></a> -<span class="sourceLineNo">374</span> * @throws IOException<a name="line.374"></a> -<span class="sourceLineNo">375</span> */<a name="line.375"></a> -<span class="sourceLineNo">376</span> protected void seekScanners(List<? extends KeyValueScanner> scanners,<a name="line.376"></a> -<span class="sourceLineNo">377</span> Cell seekKey, boolean isLazy, boolean isParallelSeek)<a name="line.377"></a> -<span class="sourceLineNo">378</span> throws IOException {<a name="line.378"></a> -<span class="sourceLineNo">379</span> // Seek all scanners to the start of the Row (or if the exact matching row<a name="line.379"></a> -<span class="sourceLineNo">380</span> // key does not exist, then to the start of the next matching Row).<a name="line.380"></a> -<span class="sourceLineNo">381</span> // Always check bloom filter to optimize the top row seek for delete<a name="line.381"></a> -<span class="sourceLineNo">382</span> // family marker.<a name="line.382"></a> -<span class="sourceLineNo">383</span> if (isLazy) {<a name="line.383"></a> -<span class="sourceLineNo">384</span> for (KeyValueScanner scanner : scanners) {<a name="line.384"></a> -<span class="sourceLineNo">385</span> scanner.requestSeek(seekKey, false, true);<a name="line.385"></a> -<span class="sourceLineNo">386</span> }<a name="line.386"></a> -<span class="sourceLineNo">387</span> } else {<a name="line.387"></a> -<span class="sourceLineNo">388</span> if (!isParallelSeek) {<a name="line.388"></a> -<span class="sourceLineNo">389</span> long totalScannersSoughtBytes = 0;<a name="line.389"></a> -<span class="sourceLineNo">390</span> for (KeyValueScanner scanner : scanners) {<a name="line.390"></a> -<span class="sourceLineNo">391</span> if (matcher.isUserScan() && totalScannersSoughtBytes >= maxRowSize) {<a name="line.391"></a> -<span class="sourceLineNo">392</span> throw new RowTooBigException("Max row size allowed: " + maxRowSize<a name="line.392"></a> -<span class="sourceLineNo">393</span> + ", but row is bigger than that");<a name="line.393"></a> -<span class="sourceLineNo">394</span> }<a name="line.394"></a> -<span class="sourceLineNo">395</span> scanner.seek(seekKey);<a name="line.395"></a> -<span class="sourceLineNo">396</span> Cell c = scanner.peek();<a name="line.396"></a> -<span class="sourceLineNo">397</span> if (c != null) {<a name="line.397"></a> -<span class="sourceLineNo">398</span> totalScannersSoughtBytes += CellUtil.estimatedSerializedSizeOf(c);<a name="line.398"></a> -<span class="sourceLineNo">399</span> }<a name="line.399"></a> -<span class="sourceLineNo">400</span> }<a name="line.400"></a> -<span class="sourceLineNo">401</span> } else {<a name="line.401"></a> -<span class="sourceLineNo">402</span> parallelSeek(scanners, seekKey);<a name="line.402"></a> -<span class="sourceLineNo">403</span> }<a name="line.403"></a> -<span class="sourceLineNo">404</span> }<a name="line.404"></a> -<span class="sourceLineNo">405</span> }<a name="line.405"></a> -<span class="sourceLineNo">406</span><a name="line.406"></a> -<span class="sourceLineNo">407</span> protected void resetKVHeap(List<? extends KeyValueScanner> scanners,<a name="line.407"></a> -<span class="sourceLineNo">408</span> CellComparator comparator) throws IOException {<a name="line.408"></a> -<span class="sourceLineNo">409</span> // Combine all seeked scanners with a heap<a name="line.409"></a> -<span class="sourceLineNo">410</span> heap = new KeyValueHeap(scanners, comparator);<a name="line.410"></a> -<span class="sourceLineNo">411</span> }<a name="line.411"></a> -<span class="sourceLineNo">412</span><a name="line.412"></a> -<span class="sourceLineNo">413</span> /**<a name="line.413"></a> -<span class="sourceLineNo">414</span> * Filters the given list of scanners using Bloom filter, time range, and<a name="line.414"></a> -<span class="sourceLineNo">415</span> * TTL.<a name="line.415"></a> -<span class="sourceLineNo">416</span> */<a name="line.416"></a> -<span class="sourceLineNo">417</span> protected List<KeyValueScanner> selectScannersFrom(<a name="line.417"></a> -<span class="sourceLineNo">418</span> final List<? extends KeyValueScanner> allScanners) {<a name="line.418"></a> -<span class="sourceLineNo">419</span> boolean memOnly;<a name="line.419"></a> -<span class="sourceLineNo">420</span> boolean filesOnly;<a name="line.420"></a> -<span class="sourceLineNo">421</span> if (scan instanceof InternalScan) {<a name="line.421"></a> -<span class="sourceLineNo">422</span> InternalScan iscan = (InternalScan)scan;<a name="line.422"></a> -<span class="sourceLineNo">423</span> memOnly = iscan.isCheckOnlyMemStore();<a name="line.423"></a> -<span class="sourceLineNo">424</span> filesOnly = iscan.isCheckOnlyStoreFiles();<a name="line.424"></a> -<span class="sourceLineNo">425</span> } else {<a name="line.425"></a> -<span class="sourceLineNo">426</span> memOnly = false;<a name="line.426"></a> -<span class="sourceLineNo">427</span> filesOnly = false;<a name="line.427"></a> -<span class="sourceLineNo">428</span> }<a name="line.428"></a> -<span class="sourceLineNo">429</span><a name="line.429"></a> -<span class="sourceLineNo">430</span> List<KeyValueScanner> scanners = new ArrayList<>(allScanners.size());<a name="line.430"></a> -<span class="sourceLineNo">431</span><a name="line.431"></a> -<span class="sourceLineNo">432</span> // We can only exclude store files based on TTL if minVersions is set to 0.<a name="line.432"></a> -<span class="sourceLineNo">433</span> // Otherwise, we might have to return KVs that have technically expired.<a name="line.433"></a> -<span class="sourceLineNo">434</span> long expiredTimestampCutoff = minVersions == 0 ? oldestUnexpiredTS: Long.MIN_VALUE;<a name="line.434"></a> -<span class="sourceLineNo">435</span><a name="line.435"></a> -<span class="sourceLineNo">436</span> // include only those scan files which pass all filters<a name="line.436"></a> -<span class="sourceLineNo">437</span> for (KeyValueScanner kvs : allScanners) {<a name="line.437"></a> -<span class="sourceLineNo">438</span> boolean isFile = kvs.isFileScanner();<a name="line.438"></a> -<span class="sourceLineNo">439</span> if ((!isFile && filesOnly) || (isFile && memOnly)) {<a name="line.439"></a> -<span class="sourceLineNo">440</span> continue;<a name="line.440"></a> -<span class="sourceLineNo">441</span> }<a name="line.441"></a> -<span class="sourceLineNo">442</span><a name="line.442"></a> -<span class="sourceLineNo">443</span> if (kvs.shouldUseScanner(scan, store, expiredTimestampCutoff)) {<a name="line.443"></a> -<span class="sourceLineNo">444</span> scanners.add(kvs);<a name="line.444"></a> -<span class="sourceLineNo">445</span> } else {<a name="line.445"></a> -<span class="sourceLineNo">446</span> kvs.close();<a name="line.446"></a> -<span class="sourceLineNo">447</span> }<a name="line.447"></a> -<span class="sourceLineNo">448</span> }<a name="line.448"></a> -<span class="sourceLineNo">449</span> return scanners;<a name="line.449"></a> -<span class="sourceLineNo">450</span> }<a name="line.450"></a> +<span class="sourceLineNo">141</span> // A flag whether use pread for scan<a name="line.141"></a> +<span class="sourceLineNo">142</span> // it maybe changed if we use Scan.ReadType.DEFAULT and we have read lots of data.<a name="line.142"></a> +<span class="sourceLineNo">143</span> private boolean scanUsePread;<a name="line.143"></a> +<span class="sourceLineNo">144</span> // Indicates whether there was flush during the course of the scan<a name="line.144"></a> +<span class="sourceLineNo">145</span> private volatile boolean flushed = false;<a name="line.145"></a> +<span class="sourceLineNo">146</span> // generally we get one file from a flush<a name="line.146"></a> +<span class="sourceLineNo">147</span> private final List<StoreFile> flushedStoreFiles = new ArrayList<>(1);<a name="line.147"></a> +<span class="sourceLineNo">148</span> // The current list of scanners<a name="line.148"></a> +<span class="sourceLineNo">149</span> private final List<KeyValueScanner> currentScanners = new ArrayList<>();<a name="line.149"></a> +<span class="sourceLineNo">150</span> // flush update lock<a name="line.150"></a> +<span class="sourceLineNo">151</span> private final ReentrantLock flushLock = new ReentrantLock();<a name="line.151"></a> +<span class="sourceLineNo">152</span><a name="line.152"></a> +<span class="sourceLineNo">153</span> protected final long readPt;<a name="line.153"></a> +<span class="sourceLineNo">154</span><a name="line.154"></a> +<span class="sourceLineNo">155</span> // used by the injection framework to test race between StoreScanner construction and compaction<a name="line.155"></a> +<span class="sourceLineNo">156</span> enum StoreScannerCompactionRace {<a name="line.156"></a> +<span class="sourceLineNo">157</span> BEFORE_SEEK,<a name="line.157"></a> +<span class="sourceLineNo">158</span> AFTER_SEEK,<a name="line.158"></a> +<span class="sourceLineNo">159</span> COMPACT_COMPLETE<a name="line.159"></a> +<span class="sourceLineNo">160</span> }<a name="line.160"></a> +<span class="sourceLineNo">161</span><a name="line.161"></a> +<span class="sourceLineNo">162</span> /** An internal constructor. */<a name="line.162"></a> +<span class="sourceLineNo">163</span> protected StoreScanner(Store store, Scan scan, final ScanInfo scanInfo,<a name="line.163"></a> +<span class="sourceLineNo">164</span> final NavigableSet<byte[]> columns, long readPt, boolean cacheBlocks) {<a name="line.164"></a> +<span class="sourceLineNo">165</span> this.readPt = readPt;<a name="line.165"></a> +<span class="sourceLineNo">166</span> this.store = store;<a name="line.166"></a> +<span class="sourceLineNo">167</span> this.cacheBlocks = cacheBlocks;<a name="line.167"></a> +<span class="sourceLineNo">168</span> get = scan.isGetScan();<a name="line.168"></a> +<span class="sourceLineNo">169</span> int numCol = columns == null ? 0 : columns.size();<a name="line.169"></a> +<span class="sourceLineNo">170</span> explicitColumnQuery = numCol > 0;<a name="line.170"></a> +<span class="sourceLineNo">171</span> this.scan = scan;<a name="line.171"></a> +<span class="sourceLineNo">172</span> this.now = EnvironmentEdgeManager.currentTime();<a name="line.172"></a> +<span class="sourceLineNo">173</span> this.oldestUnexpiredTS = scan.isRaw() ? 0L : now - scanInfo.getTtl();<a name="line.173"></a> +<span class="sourceLineNo">174</span> this.minVersions = scanInfo.getMinVersions();<a name="line.174"></a> +<span class="sourceLineNo">175</span><a name="line.175"></a> +<span class="sourceLineNo">176</span> // We look up row-column Bloom filters for multi-column queries as part of<a name="line.176"></a> +<span class="sourceLineNo">177</span> // the seek operation. However, we also look the row-column Bloom filter<a name="line.177"></a> +<span class="sourceLineNo">178</span> // for multi-row (non-"get") scans because this is not done in<a name="line.178"></a> +<span class="sourceLineNo">179</span> // StoreFile.passesBloomFilter(Scan, SortedSet<byte[]>).<a name="line.179"></a> +<span class="sourceLineNo">180</span> this.useRowColBloom = numCol > 1 || (!get && numCol == 1);<a name="line.180"></a> +<span class="sourceLineNo">181</span><a name="line.181"></a> +<span class="sourceLineNo">182</span> this.maxRowSize = scanInfo.getTableMaxRowSize();<a name="line.182"></a> +<span class="sourceLineNo">183</span> if (get) {<a name="line.183"></a> +<span class="sourceLineNo">184</span> this.readType = Scan.ReadType.PREAD;<a name="line.184"></a> +<span class="sourceLineNo">185</span> this.scanUsePread = true;<a name="line.185"></a> +<span class="sourceLineNo">186</span> } else {<a name="line.186"></a> +<span class="sourceLineNo">187</span> if (scan.getReadType() == Scan.ReadType.DEFAULT) {<a name="line.187"></a> +<span class="sourceLineNo">188</span> this.readType = scanInfo.isUsePread() ? Scan.ReadType.PREAD : Scan.ReadType.DEFAULT;<a name="line.188"></a> +<span class="sourceLineNo">189</span> } else {<a name="line.189"></a> +<span class="sourceLineNo">190</span> this.readType = scan.getReadType();<a name="line.190"></a> +<span class="sourceLineNo">191</span> }<a name="line.191"></a> +<span class="sourceLineNo">192</span> // Always start with pread unless user specific stream. Will change to stream later if<a name="line.192"></a> +<span class="sourceLineNo">193</span> // readType is default if the scan keeps running for a long time.<a name="line.193"></a> +<span class="sourceLineNo">194</span> this.scanUsePread = this.readType != Scan.ReadType.STREAM;<a name="line.194"></a> +<span class="sourceLineNo">195</span> }<a name="line.195"></a> +<span class="sourceLineNo">196</span> this.preadMaxBytes = scanInfo.getPreadMaxBytes();<a name="line.196"></a> +<span class="sourceLineNo">197</span> this.cellsPerHeartbeatCheck = scanInfo.getCellsPerTimeoutCheck();<a name="line.197"></a> +<span class="sourceLineNo">198</span> // Parallel seeking is on if the config allows and more there is more than one store file.<a name="line.198"></a> +<span class="sourceLineNo">199</span> if (this.store != null && this.store.getStorefilesCount() > 1) {<a name="line.199"></a> +<span class="sourceLineNo">200</span> RegionServerServices rsService = ((HStore) store).getHRegion().getRegionServerServices();<a name="line.200"></a> +<span class="sourceLineNo">201</span> if (rsService != null && scanInfo.isParallelSeekEnabled()) {<a name="line.201"></a> +<span class="sourceLineNo">202</span> this.parallelSeekEnabled = true;<a name="line.202"></a> +<span class="sourceLineNo">203</span> this.executor = rsService.getExecutorService();<a name="line.203"></a> +<span class="sourceLineNo">204</span> }<a name="line.204"></a> +<span class="sourceLineNo">205</span> }<a name="line.205"></a> +<span class="sourceLineNo">206</span> }<a name="line.206"></a> +<span class="sourceLineNo">207</span><a name="line.207"></a> +<span class="sourceLineNo">208</span> private void addCurrentScanners(List<? extends KeyValueScanner> scanners) {<a name="line.208"></a> +<span class="sourceLineNo">209</span> this.currentScanners.addAll(scanners);<a name="line.209"></a> +<span class="sourceLineNo">210</span> }<a name="line.210"></a> +<span class="sourceLineNo">211</span><a name="line.211"></a> +<span class="sourceLineNo">212</span> /**<a name="line.212"></a> +<span class="sourceLineNo">213</span> * Opens a scanner across memstore, snapshot, and all StoreFiles. Assumes we<a name="line.213"></a> +<span class="sourceLineNo">214</span> * are not in a compaction.<a name="line.214"></a> +<span class="sourceLineNo">215</span> *<a name="line.215"></a> +<span class="sourceLineNo">216</span> * @param store who we scan<a name="line.216"></a> +<span class="sourceLineNo">217</span> * @param scan the spec<a name="line.217"></a> +<span class="sourceLineNo">218</span> * @param columns which columns we are scanning<a name="line.218"></a> +<span class="sourceLineNo">219</span> * @throws IOException<a name="line.219"></a> +<span class="sourceLineNo">220</span> */<a name="line.220"></a> +<span class="sourceLineNo">221</span> public StoreScanner(Store store, ScanInfo scanInfo, Scan scan, final NavigableSet<byte[]> columns,<a name="line.221"></a> +<span class="sourceLineNo">222</span> long readPt)<a name="line.222"></a> +<span class="sourceLineNo">223</span> throws IOException {<a name="line.223"></a> +<span class="sourceLineNo">224</span> this(store, scan, scanInfo, columns, readPt, scan.getCacheBlocks());<a name="line.224"></a> +<span class="sourceLineNo">225</span> if (columns != null && scan.isRaw()) {<a name="line.225"></a> +<span class="sourceLineNo">226</span> throw new DoNotRetryIOException("Cannot specify any column for a raw scan");<a name="line.226"></a> +<span class="sourceLineNo">227</span> }<a name="line.227"></a> +<span class="sourceLineNo">228</span> matcher = UserScanQueryMatcher.create(scan, scanInfo, columns, oldestUnexpiredTS, now,<a name="line.228"></a> +<span class="sourceLineNo">229</span> store.getCoprocessorHost());<a name="line.229"></a> +<span class="sourceLineNo">230</span><a name="line.230"></a> +<span class="sourceLineNo">231</span> this.store.addChangedReaderObserver(this);<a name="line.231"></a> +<span class="sourceLineNo">232</span><a name="line.232"></a> +<span class="sourceLineNo">233</span> try {<a name="line.233"></a> +<span class="sourceLineNo">234</span> // Pass columns to try to filter out unnecessary StoreFiles.<a name="line.234"></a> +<span class="sourceLineNo">235</span> List<KeyValueScanner> scanners = getScannersNoCompaction();<a name="line.235"></a> +<span class="sourceLineNo">236</span><a name="line.236"></a> +<span class="sourceLineNo">237</span> // Seek all scanners to the start of the Row (or if the exact matching row<a name="line.237"></a> +<span class="sourceLineNo">238</span> // key does not exist, then to the start of the next matching Row).<a name="line.238"></a> +<span class="sourceLineNo">239</span> // Always check bloom filter to optimize the top row seek for delete<a name="line.239"></a> +<span class="sourceLineNo">240</span> // family marker.<a name="line.240"></a> +<span class="sourceLineNo">241</span> seekScanners(scanners, matcher.getStartKey(), explicitColumnQuery && lazySeekEnabledGlobally,<a name="line.241"></a> +<span class="sourceLineNo">242</span> parallelSeekEnabled);<a name="line.242"></a> +<span class="sourceLineNo">243</span><a name="line.243"></a> +<span class="sourceLineNo">244</span> // set storeLimit<a name="line.244"></a> +<span class="sourceLineNo">245</span> this.storeLimit = scan.getMaxResultsPerColumnFamily();<a name="line.245"></a> +<span class="sourceLineNo">246</span><a name="line.246"></a> +<span class="sourceLineNo">247</span> // set rowOffset<a name="line.247"></a> +<span class="sourceLineNo">248</span> this.storeOffset = scan.getRowOffsetPerColumnFamily();<a name="line.248"></a> +<span class="sourceLineNo">249</span> addCurrentScanners(scanners);<a name="line.249"></a> +<span class="sourceLineNo">250</span> // Combine all seeked scanners with a heap<a name="line.250"></a> +<span class="sourceLineNo">251</span> resetKVHeap(scanners, store.getComparator());<a name="line.251"></a> +<span class="sourceLineNo">252</span> } catch (IOException e) {<a name="line.252"></a> +<span class="sourceLineNo">253</span> // remove us from the HStore#changedReaderObservers here or we'll have no chance to<a name="line.253"></a> +<span class="sourceLineNo">254</span> // and might cause memory leak<a name="line.254"></a> +<span class="sourceLineNo">255</span> this.store.deleteChangedReaderObserver(this);<a name="line.255"></a> +<span class="sourceLineNo">256</span> throw e;<a name="line.256"></a> +<span class="sourceLineNo">257</span> }<a name="line.257"></a> +<span class="sourceLineNo">258</span> }<a name="line.258"></a> +<span class="sourceLineNo">259</span><a name="line.259"></a> +<span class="sourceLineNo">260</span> /**<a name="line.260"></a> +<span class="sourceLineNo">261</span> * Used for compactions.<p><a name="line.261"></a> +<span class="sourceLineNo">262</span> *<a name="line.262"></a> +<span class="sourceLineNo">263</span> * Opens a scanner across specified StoreFiles.<a name="line.263"></a> +<span class="sourceLineNo">264</span> * @param store who we scan<a name="line.264"></a> +<span class="sourceLineNo">265</span> * @param scan the spec<a name="line.265"></a> +<span class="sourceLineNo">266</span> * @param scanners ancillary scanners<a name="line.266"></a> +<span class="sourceLineNo">267</span> * @param smallestReadPoint the readPoint that we should use for tracking<a name="line.267"></a> +<span class="sourceLineNo">268</span> * versions<a name="line.268"></a> +<span class="sourceLineNo">269</span> */<a name="line.269"></a> +<span class="sourceLineNo">270</span> public StoreScanner(Store store, ScanInfo scanInfo, Scan scan,<a name="line.270"></a> +<span class="sourceLineNo">271</span> List<? extends KeyValueScanner> scanners, ScanType scanType,<a name="line.271"></a> +<span class="sourceLineNo">272</span> long smallestReadPoint, long earliestPutTs) throws IOException {<a name="line.272"></a> +<span class="sourceLineNo">273</span> this(store, scanInfo, scan, scanners, scanType, smallestReadPoint, earliestPutTs, null, null);<a name="line.273"></a> +<span class="sourceLineNo">274</span> }<a name="line.274"></a> +<span class="sourceLineNo">275</span><a name="line.275"></a> +<span class="sourceLineNo">276</span> /**<a name="line.276"></a> +<span class="sourceLineNo">277</span> * Used for compactions that drop deletes from a limited range of rows.<p><a name="line.277"></a> +<span class="sourceLineNo">278</span> *<a name="line.278"></a> +<span class="sourceLineNo">279</span> * Opens a scanner across specified StoreFiles.<a name="line.279"></a> +<span class="sourceLineNo">280</span> * @param store who we scan<a name="line.280"></a> +<span class="sourceLineNo">281</span> * @param scan the spec<a name="line.281"></a> +<span class="sourceLineNo">282</span> * @param scanners ancillary scanners<a name="line.282"></a> +<span class="sourceLineNo">283</span> * @param smallestReadPoint the readPoint that we should use for tracking versions<a name="line.283"></a> +<span class="sourceLineNo">284</span> * @param dropDeletesFromRow The inclusive left bound of the range; can be EMPTY_START_ROW.<a name="line.284"></a> +<span class="sourceLineNo">285</span> * @param dropDeletesToRow The exclusive right bound of the range; can be EMPTY_END_ROW.<a name="line.285"></a> +<span class="sourceLineNo">286</span> */<a name="line.286"></a> +<span class="sourceLineNo">287</span> public StoreScanner(Store store, ScanInfo scanInfo, Scan scan,<a name="line.287"></a> +<span class="sourceLineNo">288</span> List<? extends KeyValueScanner> scanners, long smallestReadPoint, long earliestPutTs,<a name="line.288"></a> +<span class="sourceLineNo">289</span> byte[] dropDeletesFromRow, byte[] dropDeletesToRow) throws IOException {<a name="line.289"></a> +<span class="sourceLineNo">290</span> this(store, scanInfo, scan, scanners, ScanType.COMPACT_RETAIN_DELETES, smallestReadPoint,<a name="line.290"></a> +<span class="sourceLineNo">291</span> earliestPutTs, dropDeletesFromRow, dropDeletesToRow);<a name="line.291"></a> +<span class="sourceLineNo">292</span> }<a name="line.292"></a> +<span class="sourceLineNo">293</span><a name="line.293"></a> +<span class="sourceLineNo">294</span> private StoreScanner(Store store, ScanInfo scanInfo, Scan scan,<a name="line.294"></a> +<span class="sourceLineNo">295</span> List<? extends KeyValueScanner> scanners, ScanType scanType, long smallestReadPoint,<a name="line.295"></a> +<span class="sourceLineNo">296</span> long earliestPutTs, byte[] dropDeletesFromRow, byte[] dropDeletesToRow) throws IOException {<a name="line.296"></a> +<span class="sourceLineNo">297</span> this(store, scan, scanInfo, null,<a name="line.297"></a> +<span class="sourceLineNo">298</span> ((HStore) store).getHRegion().getReadPoint(IsolationLevel.READ_COMMITTED), false);<a name="line.298"></a> +<span class="sourceLineNo">299</span> if (scan.hasFilter() || (scan.getStartRow() != null && scan.getStartRow().length > 0)<a name="line.299"></a> +<span class="sourceLineNo">300</span> || (scan.getStopRow() != null && scan.getStopRow().length > 0)<a name="line.300"></a> +<span class="sourceLineNo">301</span> || !scan.getTimeRange().isAllTime()) {<a name="line.301"></a> +<span class="sourceLineNo">302</span> // use legacy query matcher since we do not consider the scan object in our code. Only used to<a name="line.302"></a> +<span class="sourceLineNo">303</span> // keep compatibility for coprocessor.<a name="line.303"></a> +<span class="sourceLineNo">304</span> matcher = LegacyScanQueryMatcher.create(scan, scanInfo, null, scanType, smallestReadPoint,<a name="line.304"></a> +<span class="sourceLineNo">305</span> earliestPutTs, oldestUnexpiredTS, now, dropDeletesFromRow, dropDeletesToRow,<a name="line.305"></a> +<span class="sourceLineNo">306</span> store.getCoprocessorHost());<a name="line.306"></a> +<span class="sourceLineNo">307</span> } else {<a name="line.307"></a> +<span class="sourceLineNo">308</span> matcher = CompactionScanQueryMatcher.create(scanInfo, scanType, smallestReadPoint,<a name="line.308"></a> +<span class="sourceLineNo">309</span> earliestPutTs, oldestUnexpiredTS, now, dropDeletesFromRow, dropDeletesToRow,<a name="line.309"></a> +<span class="sourceLineNo">310</span> store.getCoprocessorHost());<a name="line.310"></a> +<span class="sourceLineNo">311</span> }<a name="line.311"></a> +<span class="sourceLineNo">312</span><a name="line.312"></a> +<span class="sourceLineNo">313</span> // Filter the list of scanners using Bloom filters, time range, TTL, etc.<a name="line.313"></a> +<span class="sourceLineNo">314</span> scanners = selectScannersFrom(scanners);<a name="line.314"></a> +<span class="sourceLineNo">315</span><a name="line.315"></a> +<span class="sourceLineNo">316</span> // Seek all scanners to the initial key<a name="line.316"></a> +<span class="sourceLineNo">317</span> seekScanners(scanners, matcher.getStartKey(), false, parallelSeekEnabled);<a name="line.317"></a> +<span class="sourceLineNo">318</span> addCurrentScanners(scanners);<a name="line.318"></a> +<span class="sourceLineNo">319</span> // Combine all seeked scanners with a heap<a name="line.319"></a> +<span class="sourceLineNo">320</span> resetKVHeap(scanners, store.getComparator());<a name="line.320"></a> +<span class="sourceLineNo">321</span> }<a name="line.321"></a> +<span class="sourceLineNo">322</span><a name="line.322"></a> +<span class="sourceLineNo">323</span> @VisibleForTesting<a name="line.323"></a> +<span class="sourceLineNo">324</span> StoreScanner(final Scan scan, ScanInfo scanInfo,<a name="line.324"></a> +<span class="sourceLineNo">325</span> ScanType scanType, final NavigableSet<byte[]> columns,<a name="line.325"></a> +<span class="sourceLineNo">326</span> final List<? extends KeyValueScanner> scanners) throws IOException {<a name="line.326"></a> +<span class="sourceLineNo">327</span> this(scan, scanInfo, scanType, columns, scanners,<a name="line.327"></a> +<span class="sourceLineNo">328</span> HConstants.LATEST_TIMESTAMP,<a name="line.328"></a> +<span class="sourceLineNo">329</span> // 0 is passed as readpoint because the test bypasses Store<a name="line.329"></a> +<span class="sourceLineNo">330</span> 0);<a name="line.330"></a> +<span class="sourceLineNo">331</span> }<a name="line.331"></a> +<span class="sourceLineNo">332</span><a name="line.332"></a> +<span class="sourceLineNo">333</span> @VisibleForTesting<a name="line.333"></a> +<span class="sourceLineNo">334</span> StoreScanner(final Scan scan, ScanInfo scanInfo,<a name="line.334"></a> +<span class="sourceLineNo">335</span> ScanType scanType, final NavigableSet<byte[]> columns,<a name="line.335"></a> +<span class="sourceLineNo">336</span> final List<? extends KeyValueScanner> scanners, long earliestPutTs)<a name="line.336"></a> +<span class="sourceLineNo">337</span> throws IOException {<a name="line.337"></a> +<span class="sourceLineNo">338</span> this(scan, scanInfo, scanType, columns, scanners, earliestPutTs,<a name="line.338"></a> +<span class="sourceLineNo">339</span> // 0 is passed as readpoint because the test bypasses Store<a name="line.339"></a> +<span class="sourceLineNo">340</span> 0);<a name="line.340"></a> +<span class="sourceLineNo">341</span> }<a name="line.341"></a> +<span class="sourceLineNo">342</span><a name="line.342"></a> +<span class="sourceLineNo">343</span> public StoreScanner(final Scan scan, ScanInfo scanInfo, ScanType scanType,<a name="line.343"></a> +<span class="sourceLineNo">344</span> final NavigableSet<byte[]> columns, final List<? extends KeyValueScanner> scanners, long earliestPutTs,<a name="line.344"></a> +<span class="sourceLineNo">345</span> long readPt) throws IOException {<a name="line.345"></a> +<span class="sourceLineNo">346</span> this(null, scan, scanInfo, columns, readPt,<a name="line.346"></a> +<span class="sourceLineNo">347</span> scanType == ScanType.USER_SCAN ? scan.getCacheBlocks() : false);<a name="line.347"></a> +<span class="sourceLineNo">348</span> if (scanType == ScanType.USER_SCAN) {<a name="line.348"></a> +<span class="sourceLineNo">349</span> this.matcher = UserScanQueryMatcher.create(scan, scanInfo, columns, oldestUnexpiredTS, now,<a name="line.349"></a> +<span class="sourceLineNo">350</span> null);<a name="line.350"></a> +<span class="sourceLineNo">351</span> } else {<a name="line.351"></a> +<span class="sourceLineNo">352</span> if (scan.hasFilter() || (scan.getStartRow() != null && scan.getStartRow().length > 0)<a name="line.352"></a> +<span class="sourceLineNo">353</span> || (scan.getStopRow() != null && scan.getStopRow().length > 0)<a name="line.353"></a> +<span class="sourceLineNo">354</span> || !scan.getTimeRange().isAllTime() || columns != null) {<a name="line.354"></a> +<span class="sourceLineNo">355</span> // use legacy query matcher since we do not consider the scan object in our code. Only used<a name="line.355"></a> +<span class="sourceLineNo">356</span> // to keep compatibility for coprocessor.<a name="line.356"></a> +<span class="sourceLineNo">357</span> matcher = LegacyScanQueryMatcher.create(scan, scanInfo, columns, scanType, Long.MAX_VALUE,<a name="line.357"></a> +<span class="sourceLineNo">358</span> earliestPutTs, oldestUnexpiredTS, now, null, null, store.getCoprocessorHost());<a name="line.358"></a> +<span class="sourceLineNo">359</span> } else {<a name="line.359"></a> +<span class="sourceLineNo">360</span> this.matcher = CompactionScanQueryMatcher.create(scanInfo, scanType, Long.MAX_VALUE,<a name="line.360"></a> +<span class="sourceLineNo">361</span> earliestPutTs, oldestUnexpiredTS, now, null, null, null);<a name="line.361"></a> +<span class="sourceLineNo">362</span> }<a name="line.362"></a> +<span class="sourceLineNo">363</span> }<a name="line.363"></a> +<span class="sourceLineNo">364</span><a name="line.364"></a> +<span class="sourceLineNo">365</span> // Seek all scanners to the initial key<a name="line.365"></a> +<span class="sourceLineNo">366</span> seekScanners(scanners, matcher.getStartKey(), false, parallelSeekEnabled);<a name="line.366"></a> +<span class="sourceLineNo">367</span> addCurrentScanners(scanners);<a name="line.367"></a> +<span class="sourceLineNo">368</span> resetKVHeap(scanners, scanInfo.getComparator());<a name="line.368"></a> +<span class="sourceLineNo">369</span> }<a name="line.369"></a> +<span class="sourceLineNo">370</span><a name="line.370"></a> +<span class="sourceLineNo">371</span> /**<a name="line.371"></a> +<span class="sourceLineNo">372</span> * Get a filtered list of scanners. Assumes we are not in a compaction.<a name="line.372"></a> +<span class="sourceLineNo">373</span> * @return list of scanners to seek<a name="line.373"></a> +<span class="sourceLineNo">374</span> */<a name="line.374"></a> +<span class="sourceLineNo">375</span> private List<KeyValueScanner> getScannersNoCompaction() throws IOException {<a name="line.375"></a> +<span class="sourceLineNo">376</span> return selectScannersFrom(<a name="line.376"></a> +<span class="sourceLineNo">377</span> store.getScanners(cacheBlocks, scanUsePread, false, matcher, scan.getStartRow(),<a name="line.377"></a> +<span class="sourceLineNo">378</span> scan.includeStartRow(), scan.getStopRow(), scan.includeStopRow(), this.readPt));<a name="line.378"></a> +<span class="sourceLineNo">379</span> }<a name="line.379"></a> +<span class="sourceLineNo">380</span><a name="line.380"></a> +<span class="sourceLineNo">381</span> /**<a name="line.381"></a> +<span class="sourceLineNo">382</span> * Seek the specified scanners with the given key<a name="line.382"></a> +<span class="sourceLineNo">383</span> * @param scanners<a name="line.383"></a> +<span class="sourceLineNo">384</span> * @param seekKey<a name="line.384"></a> +<span class="sourceLineNo">385</span> * @param isLazy true if using lazy seek<a name="line.385"></a> +<span class="sourceLineNo">386</span> * @param isParallelSeek true if using parallel seek<a name="line.386"></a> +<span class="sourceLineNo">387</span> * @throws IOException<a name="line.387"></a> +<span class="sourceLineNo">388</span> */<a name="line.388"></a> +<span class="sourceLineNo">389</span> protected void seekScanners(List<? extends KeyValueScanner> scanners,<a name="line.389"></a> +<span class="sourceLineNo">390</span> Cell seekKey, boolean isLazy, boolean isParallelSeek)<a name="line.390"></a> +<span class="sourceLineNo">391</span> throws IOException {<a name="line.391"></a> +<span class="sourceLineNo">392</span> // Seek all scanners to the start of the Row (or if the exact matching row<a name="line.392"></a> +<span class="sourceLineNo">393</span> // key does not exist, then to the start of the next matching Row).<a name="line.393"></a> +<span class="sourceLineNo">394</span> // Always check bloom filter to optimize the top row seek for delete<a name="line.394"></a> +<span class="sourceLineNo">395</span> // family marker.<a name="line.395"></a> +<span class="sourceLineNo">396</span> if (isLazy) {<a name="line.396"></a> +<span class="sourceLineNo">397</span> for (KeyValueScanner scanner : scanners) {<a name="line.397"></a> +<span class="sourceLineNo">398</span> scanner.requestSeek(seekKey, false, true);<a name="line.398"></a> +<span class="sourceLineNo">399</span> }<a name="line.399"></a> +<span class="sourceLineNo">400</span> } else {<a name="line.400"></a> +<span class="sourceLineNo">401</span> if (!isParallelSeek) {<a name="line.401"></a> +<span class="sourceLineNo">402</span> long totalScannersSoughtBytes = 0;<a name="line.402"></a> +<span class="sourceLineNo">403</span> for (KeyValueScanner scanner : scanners) {<a name="line.403"></a> +<span class="sourceLineNo">404</span> if (matcher.isUserScan() && totalScannersSoughtBytes >= maxRowSize) {<a name="line.404"></a> +<span class="sourceLineNo">405</span> throw new RowTooBigException("Max row size allowed: " + maxRowSize<a name="line.405"></a> +<span class="sourceLineNo">406</span> + ", but row is bigger than that");<a name="line.406"></a> +<span class="sourceLineNo">407</span> }<a name="line.407"></a> +<span class="sourceLineNo">408</span> scanner.seek(seekKey);<a name="line.408"></a> +<span class="sourceLineNo">409</span> Cell c = scanner.peek();<a name="line.409"></a> +<span class="sourceLineNo">410</span> if (c != null) {<a name="line.410"></a> +<span class="sourceLineNo">411</span> totalScannersSoughtBytes += CellUtil.estimatedSerializedSizeOf(c);<a name="line.411"></a> +<span class="sourceLineNo">412</span> }<a name="line.412"></a> +<span class="sourceLineNo">413</span> }<a name="line.413"></a> +<span class="sourceLineNo">414</span> } else {<a name="line.414"></a> +<span class="sourceLineNo">415</span> parallelSeek(scanners, seekKey);<a name="line.415"></a> +<span class="sourceLineNo">416</span> }<a name="line.416"></a> +<span class="sourceLineNo">417</span> }<a name="line.417"></a> +<span class="sourceLineNo">418</span> }<a name="line.418"></a> +<span class="sourceLineNo">419</span><a name="line.419"></a> +<span class="sourceLineNo">420</span> protected void resetKVHeap(List<? extends KeyValueScanner> scanners,<a name="line.420"></a> +<span class="sourceLineNo">421</span> CellComparator comparator) throws IOException {<a name="line.421"></a> +<span class="sourceLineNo">422</span> // Combine all seeked scanners with a heap<a name="line.422"></a> +<span class="sourceLineNo">423</span> heap = new KeyValueHeap(scanners, comparator);<a name="line.423"></a> +<span class="sourceLineNo">424</span> }<a name="line.424"></a> +<span class="sourceLineNo">425</span><a name="line.425"></a> +<span class="sourceLineNo">426</span> /**<a name="line.426"></a> +<span class="sourceLineNo">427</span> * Filters the given list of scanners using Bloom filter, time range, and<a name="line.427"></a> +<span class="sourceLineNo">428</span> * TTL.<a name="line.428"></a> +<span class="sourceLineNo">429</span> * <p><a name="line.429"></a> +<span class="sourceLineNo">430</span> * Will be overridden by testcase so declared as protected.<a name="line.430"></a> +<span class="sourceLineNo">431</span> */<a name="line.431"></a> +<span class="sourceLineNo">432</span> @VisibleForTesting<a name="line.432"></a> +<span class="sourceLineNo">433</span> protected List<KeyValueScanner> selectScannersFrom(<a name="line.433"></a> +<span class="sourceLineNo">434</span> final List<? extends KeyValueScanner> allScanners) {<a name="line.434"></a> +<span class="sourceLineNo">435</span> boolean memOnly;<a name="line.435"></a> +<span class="sourceLineNo">436</span> boolean filesOnly;<a name="line.436"></a> +<span class="sourceLineNo">437</span> if (scan instanceof InternalScan) {<a name="line.437"></a> +<span class="sourceLineNo">438</span> InternalScan iscan = (InternalScan)scan;<a name="line.438"></a> +<span class="sourceLineNo">439</span> memOnly = iscan.isCheckOnlyMemStore();<a name="line.439"></a> +<span class="sourceLineNo">440</span> filesOnly = iscan.isCheckOnlyStoreFiles();<a name="line.440"></a> +<span class="sourceLineNo">441</span> } else {<a name="line.441"></a> +<span class="sourceLineNo">442</span> memOnly = false;<a name="line.442"></a> +<span class="sourceLineNo">443</span> filesOnly = false;<a name="line.443"></a> +<span class="sourceLineNo">444</span> }<a name="line.444"></a> +<span class="sourceLineNo">445</span><a name="line.445"></a> +<span class="sourceLineNo">446</span> List<KeyValueScanner> scanners = new ArrayList<>(allScanners.size());<a name="line.446"></a> +<span class="sourceLineNo">447</span><a name="line.447"></a> +<span class="sourceLineNo">448</span> // We can only exclude store files based on TTL if minVersions is set to 0.<a name="line.448"></a> +<span class="sourceLineNo">449</span> // Otherwise, we might have to return KVs that have technically expired.<a name="line.449"></a> +<span class="sourceLineNo">450</span> long expiredTimestampCutoff = minVersions == 0 ? oldestUnexpiredTS: Long.MIN_VALUE;<a name="line.450"></a> <span class="sourceLineNo">451</span><a name="line.451"></a> -<span class="sourceLineNo">452</span> @Override<a name="line.452"></a> -<span class="sourceLineNo">453</span> public Cell peek() {<a name="line.453"></a> -<span class="sourceLineNo">454</span> if (this.heap == null) {<a name="line.454"></a> -<span class="sourceLineNo">455</span> return this.lastTop;<a name="line.455"></a> -<span class="sourceLineNo">456</span> }<a name="line.456"></a> -<span class="sourceLineNo">457</span> return this.heap.peek();<a name="line.457"></a> -<span class="sourceLineNo">458</span> }<a name="line.458"></a> -<span class="sourceLineNo">459</span><a name="line.459"></a> -<span class="sourceLineNo">460</span> @Override<a name="line.460"></a> -<span class="sourceLineNo">461</span> public KeyValue next() {<a name="line.461"></a> -<span class="sourceLineNo">462</span> // throw runtime exception perhaps?<a name="line.462"></a> -<span class="sourceLineNo">463</span> throw new RuntimeException("Never call StoreScanner.next()");<a name="line.463"></a> -<span class="sourceLineNo">464</span> }<a name="line.464"></a> -<span class="sourc
<TRUNCATED>
