http://git-wip-us.apache.org/repos/asf/hbase-site/blob/49852d7b/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReaderThread.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReaderThread.html b/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReaderThread.html index 4958f03..0b3870c 100644 --- a/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReaderThread.html +++ b/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReaderThread.html @@ -26,458 +26,488 @@ <span class="sourceLineNo">018</span> */<a name="line.18"></a> <span class="sourceLineNo">019</span>package org.apache.hadoop.hbase.replication.regionserver;<a name="line.19"></a> <span class="sourceLineNo">020</span><a name="line.20"></a> -<span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a> -<span class="sourceLineNo">022</span>import java.util.ArrayList;<a name="line.22"></a> -<span class="sourceLineNo">023</span>import java.util.HashMap;<a name="line.23"></a> -<span class="sourceLineNo">024</span>import java.util.List;<a name="line.24"></a> -<span class="sourceLineNo">025</span>import java.util.Map;<a name="line.25"></a> -<span class="sourceLineNo">026</span>import java.util.concurrent.BlockingQueue;<a name="line.26"></a> -<span class="sourceLineNo">027</span>import java.util.concurrent.LinkedBlockingQueue;<a name="line.27"></a> -<span class="sourceLineNo">028</span>import java.util.concurrent.PriorityBlockingQueue;<a name="line.28"></a> -<span class="sourceLineNo">029</span>import java.util.concurrent.atomic.AtomicLong;<a name="line.29"></a> -<span class="sourceLineNo">030</span><a name="line.30"></a> -<span class="sourceLineNo">031</span>import org.apache.commons.logging.Log;<a name="line.31"></a> -<span class="sourceLineNo">032</span>import org.apache.commons.logging.LogFactory;<a name="line.32"></a> -<span class="sourceLineNo">033</span>import org.apache.hadoop.conf.Configuration;<a name="line.33"></a> -<span class="sourceLineNo">034</span>import org.apache.hadoop.fs.FileSystem;<a name="line.34"></a> -<span class="sourceLineNo">035</span>import org.apache.hadoop.fs.Path;<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.CellUtil;<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.classification.InterfaceAudience;<a name="line.39"></a> -<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.classification.InterfaceStability;<a name="line.40"></a> -<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos;<a name="line.41"></a> -<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.BulkLoadDescriptor;<a name="line.42"></a> -<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.StoreDescriptor;<a name="line.43"></a> -<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.regionserver.RSRpcServices;<a name="line.44"></a> -<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.regionserver.wal.WALEdit;<a name="line.45"></a> -<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.replication.ReplicationQueueInfo;<a name="line.46"></a> -<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.replication.WALEntryFilter;<a name="line.47"></a> -<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.replication.regionserver.WALEntryStream.WALEntryStreamRuntimeException;<a name="line.48"></a> -<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.49"></a> -<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.50"></a> -<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.util.Threads;<a name="line.51"></a> -<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<a name="line.52"></a> -<span class="sourceLineNo">053</span><a name="line.53"></a> -<span class="sourceLineNo">054</span>/**<a name="line.54"></a> -<span class="sourceLineNo">055</span> * Reads and filters WAL entries, groups the filtered entries into batches, and puts the batches onto a queue<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>@InterfaceAudience.Private<a name="line.58"></a> -<span class="sourceLineNo">059</span>@InterfaceStability.Evolving<a name="line.59"></a> -<span class="sourceLineNo">060</span>public class ReplicationSourceWALReaderThread extends Thread {<a name="line.60"></a> -<span class="sourceLineNo">061</span> private static final Log LOG = LogFactory.getLog(ReplicationSourceWALReaderThread.class);<a name="line.61"></a> -<span class="sourceLineNo">062</span><a name="line.62"></a> -<span class="sourceLineNo">063</span> private PriorityBlockingQueue<Path> logQueue;<a name="line.63"></a> -<span class="sourceLineNo">064</span> private FileSystem fs;<a name="line.64"></a> -<span class="sourceLineNo">065</span> private Configuration conf;<a name="line.65"></a> -<span class="sourceLineNo">066</span> private BlockingQueue<WALEntryBatch> entryBatchQueue;<a name="line.66"></a> -<span class="sourceLineNo">067</span> // max (heap) size of each batch - multiply by number of batches in queue to get total<a name="line.67"></a> -<span class="sourceLineNo">068</span> private long replicationBatchSizeCapacity;<a name="line.68"></a> -<span class="sourceLineNo">069</span> // max count of each batch - multiply by number of batches in queue to get total<a name="line.69"></a> -<span class="sourceLineNo">070</span> private int replicationBatchCountCapacity;<a name="line.70"></a> -<span class="sourceLineNo">071</span> // position in the WAL to start reading at<a name="line.71"></a> -<span class="sourceLineNo">072</span> private long currentPosition;<a name="line.72"></a> -<span class="sourceLineNo">073</span> private WALEntryFilter filter;<a name="line.73"></a> -<span class="sourceLineNo">074</span> private long sleepForRetries;<a name="line.74"></a> -<span class="sourceLineNo">075</span> //Indicates whether this particular worker is running<a name="line.75"></a> -<span class="sourceLineNo">076</span> private boolean isReaderRunning = true;<a name="line.76"></a> -<span class="sourceLineNo">077</span> private ReplicationQueueInfo replicationQueueInfo;<a name="line.77"></a> -<span class="sourceLineNo">078</span> private int maxRetriesMultiplier;<a name="line.78"></a> -<span class="sourceLineNo">079</span> private MetricsSource metrics;<a name="line.79"></a> -<span class="sourceLineNo">080</span><a name="line.80"></a> -<span class="sourceLineNo">081</span> private AtomicLong totalBufferUsed;<a name="line.81"></a> -<span class="sourceLineNo">082</span> private long totalBufferQuota;<a name="line.82"></a> -<span class="sourceLineNo">083</span><a name="line.83"></a> -<span class="sourceLineNo">084</span> /**<a name="line.84"></a> -<span class="sourceLineNo">085</span> * Creates a reader worker for a given WAL queue. Reads WAL entries off a given queue, batches the<a name="line.85"></a> -<span class="sourceLineNo">086</span> * entries, and puts them on a batch queue.<a name="line.86"></a> -<span class="sourceLineNo">087</span> * @param manager replication manager<a name="line.87"></a> -<span class="sourceLineNo">088</span> * @param replicationQueueInfo<a name="line.88"></a> -<span class="sourceLineNo">089</span> * @param logQueue The WAL queue to read off of<a name="line.89"></a> -<span class="sourceLineNo">090</span> * @param startPosition position in the first WAL to start reading from<a name="line.90"></a> -<span class="sourceLineNo">091</span> * @param fs the files system to use<a name="line.91"></a> -<span class="sourceLineNo">092</span> * @param conf configuration to use<a name="line.92"></a> -<span class="sourceLineNo">093</span> * @param filter The filter to use while reading<a name="line.93"></a> -<span class="sourceLineNo">094</span> * @param metrics replication metrics<a name="line.94"></a> -<span class="sourceLineNo">095</span> */<a name="line.95"></a> -<span class="sourceLineNo">096</span> public ReplicationSourceWALReaderThread(ReplicationSourceManager manager,<a name="line.96"></a> -<span class="sourceLineNo">097</span> ReplicationQueueInfo replicationQueueInfo, PriorityBlockingQueue<Path> logQueue,<a name="line.97"></a> -<span class="sourceLineNo">098</span> long startPosition,<a name="line.98"></a> -<span class="sourceLineNo">099</span> FileSystem fs, Configuration conf, WALEntryFilter filter, MetricsSource metrics) {<a name="line.99"></a> -<span class="sourceLineNo">100</span> this.replicationQueueInfo = replicationQueueInfo;<a name="line.100"></a> -<span class="sourceLineNo">101</span> this.logQueue = logQueue;<a name="line.101"></a> -<span class="sourceLineNo">102</span> this.currentPosition = startPosition;<a name="line.102"></a> -<span class="sourceLineNo">103</span> this.fs = fs;<a name="line.103"></a> -<span class="sourceLineNo">104</span> this.conf = conf;<a name="line.104"></a> -<span class="sourceLineNo">105</span> this.filter = filter;<a name="line.105"></a> -<span class="sourceLineNo">106</span> this.replicationBatchSizeCapacity =<a name="line.106"></a> -<span class="sourceLineNo">107</span> this.conf.getLong("replication.source.size.capacity", 1024 * 1024 * 64);<a name="line.107"></a> -<span class="sourceLineNo">108</span> this.replicationBatchCountCapacity = this.conf.getInt("replication.source.nb.capacity", 25000);<a name="line.108"></a> -<span class="sourceLineNo">109</span> // memory used will be batchSizeCapacity * (nb.batches + 1)<a name="line.109"></a> -<span class="sourceLineNo">110</span> // the +1 is for the current thread reading before placing onto the queue<a name="line.110"></a> -<span class="sourceLineNo">111</span> int batchCount = conf.getInt("replication.source.nb.batches", 1);<a name="line.111"></a> -<span class="sourceLineNo">112</span> this.totalBufferUsed = manager.getTotalBufferUsed();<a name="line.112"></a> -<span class="sourceLineNo">113</span> this.totalBufferQuota = conf.getLong(HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_KEY,<a name="line.113"></a> -<span class="sourceLineNo">114</span> HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_DFAULT);<a name="line.114"></a> -<span class="sourceLineNo">115</span> this.sleepForRetries =<a name="line.115"></a> -<span class="sourceLineNo">116</span> this.conf.getLong("replication.source.sleepforretries", 1000); // 1 second<a name="line.116"></a> -<span class="sourceLineNo">117</span> this.maxRetriesMultiplier =<a name="line.117"></a> -<span class="sourceLineNo">118</span> this.conf.getInt("replication.source.maxretriesmultiplier", 300); // 5 minutes @ 1 sec per<a name="line.118"></a> -<span class="sourceLineNo">119</span> this.metrics = metrics;<a name="line.119"></a> -<span class="sourceLineNo">120</span> this.entryBatchQueue = new LinkedBlockingQueue<>(batchCount);<a name="line.120"></a> -<span class="sourceLineNo">121</span> LOG.info("peerClusterZnode=" + replicationQueueInfo.getPeerClusterZnode()<a name="line.121"></a> -<span class="sourceLineNo">122</span> + ", ReplicationSourceWALReaderThread : " + replicationQueueInfo.getPeerId()<a name="line.122"></a> -<span class="sourceLineNo">123</span> + " inited, replicationBatchSizeCapacity=" + replicationBatchSizeCapacity<a name="line.123"></a> -<span class="sourceLineNo">124</span> + ", replicationBatchCountCapacity=" + replicationBatchCountCapacity<a name="line.124"></a> -<span class="sourceLineNo">125</span> + ", replicationBatchQueueCapacity=" + batchCount);<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> @Override<a name="line.128"></a> -<span class="sourceLineNo">129</span> public void run() {<a name="line.129"></a> -<span class="sourceLineNo">130</span> int sleepMultiplier = 1;<a name="line.130"></a> -<span class="sourceLineNo">131</span> while (isReaderRunning()) { // we only loop back here if something fatal happened to our stream<a name="line.131"></a> -<span class="sourceLineNo">132</span> try (WALEntryStream entryStream =<a name="line.132"></a> -<span class="sourceLineNo">133</span> new WALEntryStream(logQueue, fs, conf, currentPosition, metrics)) {<a name="line.133"></a> -<span class="sourceLineNo">134</span> while (isReaderRunning()) { // loop here to keep reusing stream while we can<a name="line.134"></a> -<span class="sourceLineNo">135</span> if (!checkQuota()) {<a name="line.135"></a> -<span class="sourceLineNo">136</span> continue;<a name="line.136"></a> -<span class="sourceLineNo">137</span> }<a name="line.137"></a> -<span class="sourceLineNo">138</span> WALEntryBatch batch = null;<a name="line.138"></a> -<span class="sourceLineNo">139</span> while (entryStream.hasNext()) {<a name="line.139"></a> -<span class="sourceLineNo">140</span> if (batch == null) {<a name="line.140"></a> -<span class="sourceLineNo">141</span> batch = new WALEntryBatch(replicationBatchCountCapacity, entryStream.getCurrentPath());<a name="line.141"></a> -<span class="sourceLineNo">142</span> }<a name="line.142"></a> -<span class="sourceLineNo">143</span> Entry entry = entryStream.next();<a name="line.143"></a> -<span class="sourceLineNo">144</span> if (updateSerialReplPos(batch, entry)) {<a name="line.144"></a> -<span class="sourceLineNo">145</span> batch.lastWalPosition = entryStream.getPosition();<a name="line.145"></a> -<span class="sourceLineNo">146</span> break;<a name="line.146"></a> -<span class="sourceLineNo">147</span> }<a name="line.147"></a> -<span class="sourceLineNo">148</span> entry = filterEntry(entry);<a name="line.148"></a> -<span class="sourceLineNo">149</span> if (entry != null) {<a name="line.149"></a> -<span class="sourceLineNo">150</span> WALEdit edit = entry.getEdit();<a name="line.150"></a> -<span class="sourceLineNo">151</span> if (edit != null && !edit.isEmpty()) {<a name="line.151"></a> -<span class="sourceLineNo">152</span> long entrySize = getEntrySize(entry);<a name="line.152"></a> -<span class="sourceLineNo">153</span> batch.addEntry(entry);<a name="line.153"></a> -<span class="sourceLineNo">154</span> updateBatchStats(batch, entry, entryStream.getPosition(), entrySize);<a name="line.154"></a> -<span class="sourceLineNo">155</span> boolean totalBufferTooLarge = acquireBufferQuota(entrySize);<a name="line.155"></a> -<span class="sourceLineNo">156</span> // Stop if too many entries or too big<a name="line.156"></a> -<span class="sourceLineNo">157</span> if (totalBufferTooLarge || batch.getHeapSize() >= replicationBatchSizeCapacity<a name="line.157"></a> -<span class="sourceLineNo">158</span> || batch.getNbEntries() >= replicationBatchCountCapacity) {<a name="line.158"></a> -<span class="sourceLineNo">159</span> break;<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> }<a name="line.162"></a> -<span class="sourceLineNo">163</span> }<a name="line.163"></a> -<span class="sourceLineNo">164</span> if (batch != null && (!batch.getLastSeqIds().isEmpty() || batch.getNbEntries() > 0)) {<a name="line.164"></a> -<span class="sourceLineNo">165</span> if (LOG.isTraceEnabled()) {<a name="line.165"></a> -<span class="sourceLineNo">166</span> LOG.trace(String.format("Read %s WAL entries eligible for replication",<a name="line.166"></a> -<span class="sourceLineNo">167</span> batch.getNbEntries()));<a name="line.167"></a> -<span class="sourceLineNo">168</span> }<a name="line.168"></a> -<span class="sourceLineNo">169</span> entryBatchQueue.put(batch);<a name="line.169"></a> -<span class="sourceLineNo">170</span> sleepMultiplier = 1;<a name="line.170"></a> -<span class="sourceLineNo">171</span> } else { // got no entries and didn't advance position in WAL<a name="line.171"></a> -<span class="sourceLineNo">172</span> LOG.trace("Didn't read any new entries from WAL");<a name="line.172"></a> -<span class="sourceLineNo">173</span> if (replicationQueueInfo.isQueueRecovered()) {<a name="line.173"></a> -<span class="sourceLineNo">174</span> // we're done with queue recovery, shut ourself down<a name="line.174"></a> -<span class="sourceLineNo">175</span> setReaderRunning(false);<a name="line.175"></a> -<span class="sourceLineNo">176</span> // shuts down shipper thread immediately<a name="line.176"></a> -<span class="sourceLineNo">177</span> entryBatchQueue.put(batch != null ? batch<a name="line.177"></a> -<span class="sourceLineNo">178</span> : new WALEntryBatch(replicationBatchCountCapacity, entryStream.getCurrentPath()));<a name="line.178"></a> -<span class="sourceLineNo">179</span> } else {<a name="line.179"></a> -<span class="sourceLineNo">180</span> Thread.sleep(sleepForRetries);<a name="line.180"></a> -<span class="sourceLineNo">181</span> }<a name="line.181"></a> -<span class="sourceLineNo">182</span> }<a name="line.182"></a> -<span class="sourceLineNo">183</span> currentPosition = entryStream.getPosition();<a name="line.183"></a> -<span class="sourceLineNo">184</span> entryStream.reset(); // reuse stream<a name="line.184"></a> -<span class="sourceLineNo">185</span> }<a name="line.185"></a> -<span class="sourceLineNo">186</span> } catch (IOException | WALEntryStreamRuntimeException e) { // stream related<a name="line.186"></a> -<span class="sourceLineNo">187</span> if (sleepMultiplier < maxRetriesMultiplier) {<a name="line.187"></a> -<span class="sourceLineNo">188</span> LOG.debug("Failed to read stream of replication entries: " + e);<a name="line.188"></a> -<span class="sourceLineNo">189</span> sleepMultiplier++;<a name="line.189"></a> -<span class="sourceLineNo">190</span> } else {<a name="line.190"></a> -<span class="sourceLineNo">191</span> LOG.error("Failed to read stream of replication entries", e);<a name="line.191"></a> -<span class="sourceLineNo">192</span> }<a name="line.192"></a> -<span class="sourceLineNo">193</span> Threads.sleep(sleepForRetries * sleepMultiplier);<a name="line.193"></a> -<span class="sourceLineNo">194</span> } catch (InterruptedException e) {<a name="line.194"></a> -<span class="sourceLineNo">195</span> LOG.trace("Interrupted while sleeping between WAL reads");<a name="line.195"></a> -<span class="sourceLineNo">196</span> Thread.currentThread().interrupt();<a name="line.196"></a> -<span class="sourceLineNo">197</span> }<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> //returns false if we've already exceeded the global quota<a name="line.201"></a> -<span class="sourceLineNo">202</span> private boolean checkQuota() {<a name="line.202"></a> -<span class="sourceLineNo">203</span> // try not to go over total quota<a name="line.203"></a> -<span class="sourceLineNo">204</span> if (totalBufferUsed.get() > totalBufferQuota) {<a name="line.204"></a> -<span class="sourceLineNo">205</span> Threads.sleep(sleepForRetries);<a name="line.205"></a> -<span class="sourceLineNo">206</span> return false;<a name="line.206"></a> -<span class="sourceLineNo">207</span> }<a name="line.207"></a> -<span class="sourceLineNo">208</span> return true;<a name="line.208"></a> -<span class="sourceLineNo">209</span> }<a name="line.209"></a> -<span class="sourceLineNo">210</span><a name="line.210"></a> -<span class="sourceLineNo">211</span> private Entry filterEntry(Entry entry) {<a name="line.211"></a> -<span class="sourceLineNo">212</span> Entry filtered = filter.filter(entry);<a name="line.212"></a> -<span class="sourceLineNo">213</span> if (entry != null && filtered == null) {<a name="line.213"></a> -<span class="sourceLineNo">214</span> metrics.incrLogEditsFiltered();<a name="line.214"></a> -<span class="sourceLineNo">215</span> }<a name="line.215"></a> -<span class="sourceLineNo">216</span> return filtered;<a name="line.216"></a> -<span class="sourceLineNo">217</span> }<a name="line.217"></a> -<span class="sourceLineNo">218</span><a name="line.218"></a> -<span class="sourceLineNo">219</span> /**<a name="line.219"></a> -<span class="sourceLineNo">220</span> * @return true if we should stop reading because we're at REGION_CLOSE<a name="line.220"></a> -<span class="sourceLineNo">221</span> */<a name="line.221"></a> -<span class="sourceLineNo">222</span> private boolean updateSerialReplPos(WALEntryBatch batch, Entry entry) throws IOException {<a name="line.222"></a> -<span class="sourceLineNo">223</span> if (entry.hasSerialReplicationScope()) {<a name="line.223"></a> -<span class="sourceLineNo">224</span> String key = Bytes.toString(entry.getKey().getEncodedRegionName());<a name="line.224"></a> -<span class="sourceLineNo">225</span> batch.setLastPosition(key, entry.getKey().getSequenceId());<a name="line.225"></a> -<span class="sourceLineNo">226</span> if (!entry.getEdit().getCells().isEmpty()) {<a name="line.226"></a> -<span class="sourceLineNo">227</span> WALProtos.RegionEventDescriptor maybeEvent =<a name="line.227"></a> -<span class="sourceLineNo">228</span> WALEdit.getRegionEventDescriptor(entry.getEdit().getCells().get(0));<a name="line.228"></a> -<span class="sourceLineNo">229</span> if (maybeEvent != null && maybeEvent<a name="line.229"></a> -<span class="sourceLineNo">230</span> .getEventType() == WALProtos.RegionEventDescriptor.EventType.REGION_CLOSE) {<a name="line.230"></a> -<span class="sourceLineNo">231</span> // In serially replication, if we move a region to another RS and move it back, we may<a name="line.231"></a> -<span class="sourceLineNo">232</span> // read logs crossing two sections. We should break at REGION_CLOSE and push the first<a name="line.232"></a> -<span class="sourceLineNo">233</span> // section first in case of missing the middle section belonging to the other RS.<a name="line.233"></a> -<span class="sourceLineNo">234</span> // In a worker thread, if we can push the first log of a region, we can push all logs<a name="line.234"></a> -<span class="sourceLineNo">235</span> // in the same region without waiting until we read a close marker because next time<a name="line.235"></a> -<span class="sourceLineNo">236</span> // we read logs in this region, it must be a new section and not adjacent with this<a name="line.236"></a> -<span class="sourceLineNo">237</span> // region. Mark it negative.<a name="line.237"></a> -<span class="sourceLineNo">238</span> batch.setLastPosition(key, -entry.getKey().getSequenceId());<a name="line.238"></a> -<span class="sourceLineNo">239</span> return true;<a name="line.239"></a> -<span class="sourceLineNo">240</span> }<a name="line.240"></a> -<span class="sourceLineNo">241</span> }<a name="line.241"></a> -<span class="sourceLineNo">242</span> }<a name="line.242"></a> -<span class="sourceLineNo">243</span> return false;<a name="line.243"></a> -<span class="sourceLineNo">244</span> }<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> * Retrieves the next batch of WAL entries from the queue, waiting up to the specified time for a<a name="line.247"></a> -<span class="sourceLineNo">248</span> * batch to become available<a name="line.248"></a> -<span class="sourceLineNo">249</span> * @return A batch of entries, along with the position in the log after reading the batch<a name="line.249"></a> -<span class="sourceLineNo">250</span> * @throws InterruptedException if interrupted while waiting<a name="line.250"></a> +<span class="sourceLineNo">021</span>import java.io.EOFException;<a name="line.21"></a> +<span class="sourceLineNo">022</span>import java.io.IOException;<a name="line.22"></a> +<span class="sourceLineNo">023</span>import java.util.ArrayList;<a name="line.23"></a> +<span class="sourceLineNo">024</span>import java.util.HashMap;<a name="line.24"></a> +<span class="sourceLineNo">025</span>import java.util.List;<a name="line.25"></a> +<span class="sourceLineNo">026</span>import java.util.Map;<a name="line.26"></a> +<span class="sourceLineNo">027</span>import java.util.concurrent.BlockingQueue;<a name="line.27"></a> +<span class="sourceLineNo">028</span>import java.util.concurrent.LinkedBlockingQueue;<a name="line.28"></a> +<span class="sourceLineNo">029</span>import java.util.concurrent.PriorityBlockingQueue;<a name="line.29"></a> +<span class="sourceLineNo">030</span>import java.util.concurrent.atomic.AtomicLong;<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.conf.Configuration;<a name="line.34"></a> +<span class="sourceLineNo">035</span>import org.apache.hadoop.fs.FileSystem;<a name="line.35"></a> +<span class="sourceLineNo">036</span>import org.apache.hadoop.fs.Path;<a name="line.36"></a> +<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.Cell;<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.HConstants;<a name="line.39"></a> +<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.40"></a> +<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.classification.InterfaceStability;<a name="line.41"></a> +<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos;<a name="line.42"></a> +<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.BulkLoadDescriptor;<a name="line.43"></a> +<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.StoreDescriptor;<a name="line.44"></a> +<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.regionserver.RSRpcServices;<a name="line.45"></a> +<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.regionserver.wal.WALEdit;<a name="line.46"></a> +<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.replication.ReplicationQueueInfo;<a name="line.47"></a> +<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.replication.WALEntryFilter;<a name="line.48"></a> +<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.replication.regionserver.WALEntryStream.WALEntryStreamRuntimeException;<a name="line.49"></a> +<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.50"></a> +<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.51"></a> +<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.util.Threads;<a name="line.52"></a> +<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<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> * Reads and filters WAL entries, groups the filtered entries into batches, and puts the batches onto a queue<a name="line.56"></a> +<span class="sourceLineNo">057</span> *<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>@InterfaceStability.Evolving<a name="line.60"></a> +<span class="sourceLineNo">061</span>public class ReplicationSourceWALReaderThread extends Thread {<a name="line.61"></a> +<span class="sourceLineNo">062</span> private static final Log LOG = LogFactory.getLog(ReplicationSourceWALReaderThread.class);<a name="line.62"></a> +<span class="sourceLineNo">063</span><a name="line.63"></a> +<span class="sourceLineNo">064</span> private PriorityBlockingQueue<Path> logQueue;<a name="line.64"></a> +<span class="sourceLineNo">065</span> private FileSystem fs;<a name="line.65"></a> +<span class="sourceLineNo">066</span> private Configuration conf;<a name="line.66"></a> +<span class="sourceLineNo">067</span> private BlockingQueue<WALEntryBatch> entryBatchQueue;<a name="line.67"></a> +<span class="sourceLineNo">068</span> // max (heap) size of each batch - multiply by number of batches in queue to get total<a name="line.68"></a> +<span class="sourceLineNo">069</span> private long replicationBatchSizeCapacity;<a name="line.69"></a> +<span class="sourceLineNo">070</span> // max count of each batch - multiply by number of batches in queue to get total<a name="line.70"></a> +<span class="sourceLineNo">071</span> private int replicationBatchCountCapacity;<a name="line.71"></a> +<span class="sourceLineNo">072</span> // position in the WAL to start reading at<a name="line.72"></a> +<span class="sourceLineNo">073</span> private long currentPosition;<a name="line.73"></a> +<span class="sourceLineNo">074</span> private WALEntryFilter filter;<a name="line.74"></a> +<span class="sourceLineNo">075</span> private long sleepForRetries;<a name="line.75"></a> +<span class="sourceLineNo">076</span> //Indicates whether this particular worker is running<a name="line.76"></a> +<span class="sourceLineNo">077</span> private boolean isReaderRunning = true;<a name="line.77"></a> +<span class="sourceLineNo">078</span> private ReplicationQueueInfo replicationQueueInfo;<a name="line.78"></a> +<span class="sourceLineNo">079</span> private int maxRetriesMultiplier;<a name="line.79"></a> +<span class="sourceLineNo">080</span> private MetricsSource metrics;<a name="line.80"></a> +<span class="sourceLineNo">081</span><a name="line.81"></a> +<span class="sourceLineNo">082</span> private AtomicLong totalBufferUsed;<a name="line.82"></a> +<span class="sourceLineNo">083</span> private long totalBufferQuota;<a name="line.83"></a> +<span class="sourceLineNo">084</span><a name="line.84"></a> +<span class="sourceLineNo">085</span> /**<a name="line.85"></a> +<span class="sourceLineNo">086</span> * Creates a reader worker for a given WAL queue. Reads WAL entries off a given queue, batches the<a name="line.86"></a> +<span class="sourceLineNo">087</span> * entries, and puts them on a batch queue.<a name="line.87"></a> +<span class="sourceLineNo">088</span> * @param manager replication manager<a name="line.88"></a> +<span class="sourceLineNo">089</span> * @param replicationQueueInfo<a name="line.89"></a> +<span class="sourceLineNo">090</span> * @param logQueue The WAL queue to read off of<a name="line.90"></a> +<span class="sourceLineNo">091</span> * @param startPosition position in the first WAL to start reading from<a name="line.91"></a> +<span class="sourceLineNo">092</span> * @param fs the files system to use<a name="line.92"></a> +<span class="sourceLineNo">093</span> * @param conf configuration to use<a name="line.93"></a> +<span class="sourceLineNo">094</span> * @param filter The filter to use while reading<a name="line.94"></a> +<span class="sourceLineNo">095</span> * @param metrics replication metrics<a name="line.95"></a> +<span class="sourceLineNo">096</span> */<a name="line.96"></a> +<span class="sourceLineNo">097</span> public ReplicationSourceWALReaderThread(ReplicationSourceManager manager,<a name="line.97"></a> +<span class="sourceLineNo">098</span> ReplicationQueueInfo replicationQueueInfo, PriorityBlockingQueue<Path> logQueue,<a name="line.98"></a> +<span class="sourceLineNo">099</span> long startPosition,<a name="line.99"></a> +<span class="sourceLineNo">100</span> FileSystem fs, Configuration conf, WALEntryFilter filter, MetricsSource metrics) {<a name="line.100"></a> +<span class="sourceLineNo">101</span> this.replicationQueueInfo = replicationQueueInfo;<a name="line.101"></a> +<span class="sourceLineNo">102</span> this.logQueue = logQueue;<a name="line.102"></a> +<span class="sourceLineNo">103</span> this.currentPosition = startPosition;<a name="line.103"></a> +<span class="sourceLineNo">104</span> this.fs = fs;<a name="line.104"></a> +<span class="sourceLineNo">105</span> this.conf = conf;<a name="line.105"></a> +<span class="sourceLineNo">106</span> this.filter = filter;<a name="line.106"></a> +<span class="sourceLineNo">107</span> this.replicationBatchSizeCapacity =<a name="line.107"></a> +<span class="sourceLineNo">108</span> this.conf.getLong("replication.source.size.capacity", 1024 * 1024 * 64);<a name="line.108"></a> +<span class="sourceLineNo">109</span> this.replicationBatchCountCapacity = this.conf.getInt("replication.source.nb.capacity", 25000);<a name="line.109"></a> +<span class="sourceLineNo">110</span> // memory used will be batchSizeCapacity * (nb.batches + 1)<a name="line.110"></a> +<span class="sourceLineNo">111</span> // the +1 is for the current thread reading before placing onto the queue<a name="line.111"></a> +<span class="sourceLineNo">112</span> int batchCount = conf.getInt("replication.source.nb.batches", 1);<a name="line.112"></a> +<span class="sourceLineNo">113</span> this.totalBufferUsed = manager.getTotalBufferUsed();<a name="line.113"></a> +<span class="sourceLineNo">114</span> this.totalBufferQuota = conf.getLong(HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_KEY,<a name="line.114"></a> +<span class="sourceLineNo">115</span> HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_DFAULT);<a name="line.115"></a> +<span class="sourceLineNo">116</span> this.sleepForRetries =<a name="line.116"></a> +<span class="sourceLineNo">117</span> this.conf.getLong("replication.source.sleepforretries", 1000); // 1 second<a name="line.117"></a> +<span class="sourceLineNo">118</span> this.maxRetriesMultiplier =<a name="line.118"></a> +<span class="sourceLineNo">119</span> this.conf.getInt("replication.source.maxretriesmultiplier", 300); // 5 minutes @ 1 sec per<a name="line.119"></a> +<span class="sourceLineNo">120</span> this.metrics = metrics;<a name="line.120"></a> +<span class="sourceLineNo">121</span> this.entryBatchQueue = new LinkedBlockingQueue<>(batchCount);<a name="line.121"></a> +<span class="sourceLineNo">122</span> LOG.info("peerClusterZnode=" + replicationQueueInfo.getPeerClusterZnode()<a name="line.122"></a> +<span class="sourceLineNo">123</span> + ", ReplicationSourceWALReaderThread : " + replicationQueueInfo.getPeerId()<a name="line.123"></a> +<span class="sourceLineNo">124</span> + " inited, replicationBatchSizeCapacity=" + replicationBatchSizeCapacity<a name="line.124"></a> +<span class="sourceLineNo">125</span> + ", replicationBatchCountCapacity=" + replicationBatchCountCapacity<a name="line.125"></a> +<span class="sourceLineNo">126</span> + ", replicationBatchQueueCapacity=" + batchCount);<a name="line.126"></a> +<span class="sourceLineNo">127</span> }<a name="line.127"></a> +<span class="sourceLineNo">128</span><a name="line.128"></a> +<span class="sourceLineNo">129</span> @Override<a name="line.129"></a> +<span class="sourceLineNo">130</span> public void run() {<a name="line.130"></a> +<span class="sourceLineNo">131</span> int sleepMultiplier = 1;<a name="line.131"></a> +<span class="sourceLineNo">132</span> while (isReaderRunning()) { // we only loop back here if something fatal happened to our stream<a name="line.132"></a> +<span class="sourceLineNo">133</span> try (WALEntryStream entryStream =<a name="line.133"></a> +<span class="sourceLineNo">134</span> new WALEntryStream(logQueue, fs, conf, currentPosition, metrics)) {<a name="line.134"></a> +<span class="sourceLineNo">135</span> while (isReaderRunning()) { // loop here to keep reusing stream while we can<a name="line.135"></a> +<span class="sourceLineNo">136</span> if (!checkQuota()) {<a name="line.136"></a> +<span class="sourceLineNo">137</span> continue;<a name="line.137"></a> +<span class="sourceLineNo">138</span> }<a name="line.138"></a> +<span class="sourceLineNo">139</span> WALEntryBatch batch = null;<a name="line.139"></a> +<span class="sourceLineNo">140</span> while (entryStream.hasNext()) {<a name="line.140"></a> +<span class="sourceLineNo">141</span> if (batch == null) {<a name="line.141"></a> +<span class="sourceLineNo">142</span> batch = new WALEntryBatch(replicationBatchCountCapacity, entryStream.getCurrentPath());<a name="line.142"></a> +<span class="sourceLineNo">143</span> }<a name="line.143"></a> +<span class="sourceLineNo">144</span> Entry entry = entryStream.next();<a name="line.144"></a> +<span class="sourceLineNo">145</span> if (updateSerialReplPos(batch, entry)) {<a name="line.145"></a> +<span class="sourceLineNo">146</span> batch.lastWalPosition = entryStream.getPosition();<a name="line.146"></a> +<span class="sourceLineNo">147</span> break;<a name="line.147"></a> +<span class="sourceLineNo">148</span> }<a name="line.148"></a> +<span class="sourceLineNo">149</span> entry = filterEntry(entry);<a name="line.149"></a> +<span class="sourceLineNo">150</span> if (entry != null) {<a name="line.150"></a> +<span class="sourceLineNo">151</span> WALEdit edit = entry.getEdit();<a name="line.151"></a> +<span class="sourceLineNo">152</span> if (edit != null && !edit.isEmpty()) {<a name="line.152"></a> +<span class="sourceLineNo">153</span> long entrySize = getEntrySize(entry);<a name="line.153"></a> +<span class="sourceLineNo">154</span> batch.addEntry(entry);<a name="line.154"></a> +<span class="sourceLineNo">155</span> updateBatchStats(batch, entry, entryStream.getPosition(), entrySize);<a name="line.155"></a> +<span class="sourceLineNo">156</span> boolean totalBufferTooLarge = acquireBufferQuota(entrySize);<a name="line.156"></a> +<span class="sourceLineNo">157</span> // Stop if too many entries or too big<a name="line.157"></a> +<span class="sourceLineNo">158</span> if (totalBufferTooLarge || batch.getHeapSize() >= replicationBatchSizeCapacity<a name="line.158"></a> +<span class="sourceLineNo">159</span> || batch.getNbEntries() >= replicationBatchCountCapacity) {<a name="line.159"></a> +<span class="sourceLineNo">160</span> break;<a name="line.160"></a> +<span class="sourceLineNo">161</span> }<a name="line.161"></a> +<span class="sourceLineNo">162</span> }<a name="line.162"></a> +<span class="sourceLineNo">163</span> }<a name="line.163"></a> +<span class="sourceLineNo">164</span> }<a name="line.164"></a> +<span class="sourceLineNo">165</span> if (batch != null && (!batch.getLastSeqIds().isEmpty() || batch.getNbEntries() > 0)) {<a name="line.165"></a> +<span class="sourceLineNo">166</span> if (LOG.isTraceEnabled()) {<a name="line.166"></a> +<span class="sourceLineNo">167</span> LOG.trace(String.format("Read %s WAL entries eligible for replication",<a name="line.167"></a> +<span class="sourceLineNo">168</span> batch.getNbEntries()));<a name="line.168"></a> +<span class="sourceLineNo">169</span> }<a name="line.169"></a> +<span class="sourceLineNo">170</span> entryBatchQueue.put(batch);<a name="line.170"></a> +<span class="sourceLineNo">171</span> sleepMultiplier = 1;<a name="line.171"></a> +<span class="sourceLineNo">172</span> } else { // got no entries and didn't advance position in WAL<a name="line.172"></a> +<span class="sourceLineNo">173</span> LOG.trace("Didn't read any new entries from WAL");<a name="line.173"></a> +<span class="sourceLineNo">174</span> if (replicationQueueInfo.isQueueRecovered()) {<a name="line.174"></a> +<span class="sourceLineNo">175</span> // we're done with queue recovery, shut ourself down<a name="line.175"></a> +<span class="sourceLineNo">176</span> setReaderRunning(false);<a name="line.176"></a> +<span class="sourceLineNo">177</span> // shuts down shipper thread immediately<a name="line.177"></a> +<span class="sourceLineNo">178</span> entryBatchQueue.put(batch != null ? batch<a name="line.178"></a> +<span class="sourceLineNo">179</span> : new WALEntryBatch(replicationBatchCountCapacity, entryStream.getCurrentPath()));<a name="line.179"></a> +<span class="sourceLineNo">180</span> } else {<a name="line.180"></a> +<span class="sourceLineNo">181</span> Thread.sleep(sleepForRetries);<a name="line.181"></a> +<span class="sourceLineNo">182</span> }<a name="line.182"></a> +<span class="sourceLineNo">183</span> }<a name="line.183"></a> +<span class="sourceLineNo">184</span> currentPosition = entryStream.getPosition();<a name="line.184"></a> +<span class="sourceLineNo">185</span> entryStream.reset(); // reuse stream<a name="line.185"></a> +<span class="sourceLineNo">186</span> }<a name="line.186"></a> +<span class="sourceLineNo">187</span> } catch (IOException | WALEntryStreamRuntimeException e) { // stream related<a name="line.187"></a> +<span class="sourceLineNo">188</span> if (sleepMultiplier < maxRetriesMultiplier) {<a name="line.188"></a> +<span class="sourceLineNo">189</span> LOG.debug("Failed to read stream of replication entries: " + e);<a name="line.189"></a> +<span class="sourceLineNo">190</span> sleepMultiplier++;<a name="line.190"></a> +<span class="sourceLineNo">191</span> } else {<a name="line.191"></a> +<span class="sourceLineNo">192</span> LOG.error("Failed to read stream of replication entries", e);<a name="line.192"></a> +<span class="sourceLineNo">193</span> handleEofException(e);<a name="line.193"></a> +<span class="sourceLineNo">194</span> }<a name="line.194"></a> +<span class="sourceLineNo">195</span> Threads.sleep(sleepForRetries * sleepMultiplier);<a name="line.195"></a> +<span class="sourceLineNo">196</span> } catch (InterruptedException e) {<a name="line.196"></a> +<span class="sourceLineNo">197</span> LOG.trace("Interrupted while sleeping between WAL reads");<a name="line.197"></a> +<span class="sourceLineNo">198</span> Thread.currentThread().interrupt();<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> }<a name="line.201"></a> +<span class="sourceLineNo">202</span><a name="line.202"></a> +<span class="sourceLineNo">203</span> // if we get an EOF due to a zero-length log, and there are other logs in queue<a name="line.203"></a> +<span class="sourceLineNo">204</span> // (highly likely we've closed the current log), we've hit the max retries, and autorecovery is<a name="line.204"></a> +<span class="sourceLineNo">205</span> // enabled, then dump the log<a name="line.205"></a> +<span class="sourceLineNo">206</span> private void handleEofException(Exception e) {<a name="line.206"></a> +<span class="sourceLineNo">207</span> if (e.getCause() instanceof EOFException && logQueue.size() > 1<a name="line.207"></a> +<span class="sourceLineNo">208</span> && conf.getBoolean("replication.source.eof.autorecovery", false)) {<a name="line.208"></a> +<span class="sourceLineNo">209</span> try {<a name="line.209"></a> +<span class="sourceLineNo">210</span> if (fs.getFileStatus(logQueue.peek()).getLen() == 0) {<a name="line.210"></a> +<span class="sourceLineNo">211</span> LOG.warn("Forcing removal of 0 length log in queue: " + logQueue.peek());<a name="line.211"></a> +<span class="sourceLineNo">212</span> logQueue.remove();<a name="line.212"></a> +<span class="sourceLineNo">213</span> currentPosition = 0;<a name="line.213"></a> +<span class="sourceLineNo">214</span> }<a name="line.214"></a> +<span class="sourceLineNo">215</span> } catch (IOException ioe) {<a name="line.215"></a> +<span class="sourceLineNo">216</span> LOG.warn("Couldn't get file length information about log " + logQueue.peek());<a name="line.216"></a> +<span class="sourceLineNo">217</span> }<a name="line.217"></a> +<span class="sourceLineNo">218</span> }<a name="line.218"></a> +<span class="sourceLineNo">219</span> }<a name="line.219"></a> +<span class="sourceLineNo">220</span><a name="line.220"></a> +<span class="sourceLineNo">221</span> public Path getCurrentPath() {<a name="line.221"></a> +<span class="sourceLineNo">222</span> // if we've read some WAL entries, get the Path we read from<a name="line.222"></a> +<span class="sourceLineNo">223</span> WALEntryBatch batchQueueHead = entryBatchQueue.peek();<a name="line.223"></a> +<span class="sourceLineNo">224</span> if (batchQueueHead != null) {<a name="line.224"></a> +<span class="sourceLineNo">225</span> return batchQueueHead.lastWalPath;<a name="line.225"></a> +<span class="sourceLineNo">226</span> }<a name="line.226"></a> +<span class="sourceLineNo">227</span> // otherwise, we must be currently reading from the head of the log queue<a name="line.227"></a> +<span class="sourceLineNo">228</span> return logQueue.peek();<a name="line.228"></a> +<span class="sourceLineNo">229</span> }<a name="line.229"></a> +<span class="sourceLineNo">230</span><a name="line.230"></a> +<span class="sourceLineNo">231</span> //returns false if we've already exceeded the global quota<a name="line.231"></a> +<span class="sourceLineNo">232</span> private boolean checkQuota() {<a name="line.232"></a> +<span class="sourceLineNo">233</span> // try not to go over total quota<a name="line.233"></a> +<span class="sourceLineNo">234</span> if (totalBufferUsed.get() > totalBufferQuota) {<a name="line.234"></a> +<span class="sourceLineNo">235</span> Threads.sleep(sleepForRetries);<a name="line.235"></a> +<span class="sourceLineNo">236</span> return false;<a name="line.236"></a> +<span class="sourceLineNo">237</span> }<a name="line.237"></a> +<span class="sourceLineNo">238</span> return true;<a name="line.238"></a> +<span class="sourceLineNo">239</span> }<a name="line.239"></a> +<span class="sourceLineNo">240</span><a name="line.240"></a> +<span class="sourceLineNo">241</span> private Entry filterEntry(Entry entry) {<a name="line.241"></a> +<span class="sourceLineNo">242</span> Entry filtered = filter.filter(entry);<a name="line.242"></a> +<span class="sourceLineNo">243</span> if (entry != null && filtered == null) {<a name="line.243"></a> +<span class="sourceLineNo">244</span> metrics.incrLogEditsFiltered();<a name="line.244"></a> +<span class="sourceLineNo">245</span> }<a name="line.245"></a> +<span class="sourceLineNo">246</span> return filtered;<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> /**<a name="line.249"></a> +<span class="sourceLineNo">250</span> * @return true if we should stop reading because we're at REGION_CLOSE<a name="line.250"></a> <span class="sourceLineNo">251</span> */<a name="line.251"></a> -<span class="sourceLineNo">252</span> public WALEntryBatch take() throws InterruptedException {<a name="line.252"></a> -<span class="sourceLineNo">253</span> return entryBatchQueue.take();<a name="line.253"></a> -<span class="sourceLineNo">254</span> }<a name="line.254"></a> -<span class="sourceLineNo">255</span><a name="line.255"></a> -<span class="sourceLineNo">256</span> private long getEntrySize(Entry entry) {<a name="line.256"></a> -<span class="sourceLineNo">257</span> WALEdit edit = entry.getEdit();<a name="line.257"></a> -<span class="sourceLineNo">258</span> return edit.heapSize() + calculateTotalSizeOfStoreFiles(edit);<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> private void updateBatchStats(WALEntryBatch batch, Entry entry, long entryPosition, long entrySize) {<a name="line.261"></a> -<span class="sourceLineNo">262</span> WALEdit edit = entry.getEdit();<a name="line.262"></a> -<span class="sourceLineNo">263</span> if (edit != null && !edit.isEmpty()) {<a name="line.263"></a> -<span class="sourceLineNo">264</span> batch.incrementHeapSize(entrySize);<a name="line.264"></a> -<span class="sourceLineNo">265</span> Pair<Integer, Integer> nbRowsAndHFiles = countDistinctRowKeysAndHFiles(edit);<a name="line.265"></a> -<span class="sourceLineNo">266</span> batch.incrementNbRowKeys(nbRowsAndHFiles.getFirst());<a name="line.266"></a> -<span class="sourceLineNo">267</span> batch.incrementNbHFiles(nbRowsAndHFiles.getSecond());<a name="line.267"></a> -<span class="sourceLineNo">268</span> }<a name="line.268"></a> -<span class="sourceLineNo">269</span> batch.lastWalPosition = entryPosition;<a name="line.269"></a> -<span class="sourceLineNo">270</span> }<a name="line.270"></a> -<span class="sourceLineNo">271</span><a name="line.271"></a> -<span class="sourceLineNo">272</span> /**<a name="line.272"></a> -<span class="sourceLineNo">273</span> * Count the number of different row keys in the given edit because of mini-batching. We assume<a name="line.273"></a> -<span class="sourceLineNo">274</span> * that there's at least one Cell in the WALEdit.<a name="line.274"></a> -<span class="sourceLineNo">275</span> * @param edit edit to count row keys from<a name="line.275"></a> -<span class="sourceLineNo">276</span> * @return number of different row keys and HFiles<a name="line.276"></a> -<span class="sourceLineNo">277</span> */<a name="line.277"></a> -<span class="sourceLineNo">278</span> private Pair<Integer, Integer> countDistinctRowKeysAndHFiles(WALEdit edit) {<a name="line.278"></a> -<span class="sourceLineNo">279</span> List<Cell> cells = edit.getCells();<a name="line.279"></a> -<span class="sourceLineNo">280</span> int distinctRowKeys = 1;<a name="line.280"></a> -<span class="sourceLineNo">281</span> int totalHFileEntries = 0;<a name="line.281"></a> -<span class="sourceLineNo">282</span> Cell lastCell = cells.get(0);<a name="line.282"></a> -<span class="sourceLineNo">283</span><a name="line.283"></a> -<span class="sourceLineNo">284</span> int totalCells = edit.size();<a name="line.284"></a> -<span class="sourceLineNo">285</span> for (int i = 0; i < totalCells; i++) {<a name="line.285"></a> -<span class="sourceLineNo">286</span> // Count HFiles to be replicated<a name="line.286"></a> -<span class="sourceLineNo">287</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.287"></a> -<span class="sourceLineNo">288</span> try {<a name="line.288"></a> -<span class="sourceLineNo">289</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.289"></a> -<span class="sourceLineNo">290</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.290"></a> -<span class="sourceLineNo">291</span> int totalStores = stores.size();<a name="line.291"></a> -<span class="sourceLineNo">292</span> for (int j = 0; j < totalStores; j++) {<a name="line.292"></a> -<span class="sourceLineNo">293</span> totalHFileEntries += stores.get(j).getStoreFileList().size();<a name="line.293"></a> -<span class="sourceLineNo">294</span> }<a name="line.294"></a> -<span class="sourceLineNo">295</span> } catch (IOException e) {<a name="line.295"></a> -<span class="sourceLineNo">296</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.296"></a> -<span class="sourceLineNo">297</span> + "Then its hfiles count will not be added into metric.");<a name="line.297"></a> -<span class="sourceLineNo">298</span> }<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> if (!CellUtil.matchingRows(cells.get(i), lastCell)) {<a name="line.301"></a> -<span class="sourceLineNo">302</span> distinctRowKeys++;<a name="line.302"></a> -<span class="sourceLineNo">303</span> }<a name="line.303"></a> -<span class="sourceLineNo">304</span> lastCell = cells.get(i);<a name="line.304"></a> -<span class="sourceLineNo">305</span> }<a name="line.305"></a> -<span class="sourceLineNo">306</span><a name="line.306"></a> -<span class="sourceLineNo">307</span> Pair<Integer, Integer> result = new Pair<>(distinctRowKeys, totalHFileEntries);<a name="line.307"></a> -<span class="sourceLineNo">308</span> return result;<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> /**<a name="line.311"></a> -<span class="sourceLineNo">312</span> * Calculate the total size of all the store files<a name="line.312"></a> -<span class="sourceLineNo">313</span> * @param edit edit to count row keys from<a name="line.313"></a> -<span class="sourceLineNo">314</span> * @return the total size of the store files<a name="line.314"></a> -<span class="sourceLineNo">315</span> */<a name="line.315"></a> -<span class="sourceLineNo">316</span> private int calculateTotalSizeOfStoreFiles(WALEdit edit) {<a name="line.316"></a> -<span class="sourceLineNo">317</span> List<Cell> cells = edit.getCells();<a name="line.317"></a> -<span class="sourceLineNo">318</span> int totalStoreFilesSize = 0;<a name="line.318"></a> -<span class="sourceLineNo">319</span><a name="line.319"></a> -<span class="sourceLineNo">320</span> int totalCells = edit.size();<a name="line.320"></a> -<span class="sourceLineNo">321</span> for (int i = 0; i < totalCells; i++) {<a name="line.321"></a> -<span class="sourceLineNo">322</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.322"></a> -<span class="sourceLineNo">323</span> try {<a name="line.323"></a> -<span class="sourceLineNo">324</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.324"></a> -<span class="sourceLineNo">325</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.325"></a> -<span class="sourceLineNo">326</span> int totalStores = stores.size();<a name="line.326"></a> -<span class="sourceLineNo">327</span> for (int j = 0; j < totalStores; j++) {<a name="line.327"></a> -<span class="sourceLineNo">328</span> totalStoreFilesSize += stores.get(j).getStoreFileSizeBytes();<a name="line.328"></a> -<span class="sourceLineNo">329</span> }<a name="line.329"></a> -<span class="sourceLineNo">330</span> } catch (IOException e) {<a name="line.330"></a> -<span class="sourceLineNo">331</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.331"></a> -<span class="sourceLineNo">332</span> + "Size of HFiles part of cell will not be considered in replication "<a name="line.332"></a> -<span class="sourceLineNo">333</span> + "request size calculation.",<a name="line.333"></a> -<span class="sourceLineNo">334</span> e);<a name="line.334"></a> -<span class="sourceLineNo">335</span> }<a name="line.335"></a> -<span class="sourceLineNo">336</span> }<a name="line.336"></a> -<span class="sourceLineNo">337</span> }<a name="line.337"></a> -<span class="sourceLineNo">338</span> return totalStoreFilesSize;<a name="line.338"></a> +<span class="sourceLineNo">252</span> private boolean updateSerialReplPos(WALEntryBatch batch, Entry entry) throws IOException {<a name="line.252"></a> +<span class="sourceLineNo">253</span> if (entry.hasSerialReplicationScope()) {<a name="line.253"></a> +<span class="sourceLineNo">254</span> String key = Bytes.toString(entry.getKey().getEncodedRegionName());<a name="line.254"></a> +<span class="sourceLineNo">255</span> batch.setLastPosition(key, entry.getKey().getSequenceId());<a name="line.255"></a> +<span class="sourceLineNo">256</span> if (!entry.getEdit().getCells().isEmpty()) {<a name="line.256"></a> +<span class="sourceLineNo">257</span> WALProtos.RegionEventDescriptor maybeEvent =<a name="line.257"></a> +<span class="sourceLineNo">258</span> WALEdit.getRegionEventDescriptor(entry.getEdit().getCells().get(0));<a name="line.258"></a> +<span class="sourceLineNo">259</span> if (maybeEvent != null && maybeEvent<a name="line.259"></a> +<span class="sourceLineNo">260</span> .getEventType() == WALProtos.RegionEventDescriptor.EventType.REGION_CLOSE) {<a name="line.260"></a> +<span class="sourceLineNo">261</span> // In serially replication, if we move a region to another RS and move it back, we may<a name="line.261"></a> +<span class="sourceLineNo">262</span> // read logs crossing two sections. We should break at REGION_CLOSE and push the first<a name="line.262"></a> +<span class="sourceLineNo">263</span> // section first in case of missing the middle section belonging to the other RS.<a name="line.263"></a> +<span class="sourceLineNo">264</span> // In a worker thread, if we can push the first log of a region, we can push all logs<a name="line.264"></a> +<span class="sourceLineNo">265</span> // in the same region without waiting until we read a close marker because next time<a name="line.265"></a> +<span class="sourceLineNo">266</span> // we read logs in this region, it must be a new section and not adjacent with this<a name="line.266"></a> +<span class="sourceLineNo">267</span> // region. Mark it negative.<a name="line.267"></a> +<span class="sourceLineNo">268</span> batch.setLastPosition(key, -entry.getKey().getSequenceId());<a name="line.268"></a> +<span class="sourceLineNo">269</span> return true;<a name="line.269"></a> +<span class="sourceLineNo">270</span> }<a name="line.270"></a> +<span class="sourceLineNo">271</span> }<a name="line.271"></a> +<span class="sourceLineNo">272</span> }<a name="line.272"></a> +<span class="sourceLineNo">273</span> return false;<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> * Retrieves the next batch of WAL entries from the queue, waiting up to the specified time for a<a name="line.277"></a> +<span class="sourceLineNo">278</span> * batch to become available<a name="line.278"></a> +<span class="sourceLineNo">279</span> * @return A batch of entries, along with the position in the log after reading the batch<a name="line.279"></a> +<span class="sourceLineNo">280</span> * @throws InterruptedException if interrupted while waiting<a name="line.280"></a> +<span class="sourceLineNo">281</span> */<a name="line.281"></a> +<span class="sourceLineNo">282</span> public WALEntryBatch take() throws InterruptedException {<a name="line.282"></a> +<span class="sourceLineNo">283</span> return entryBatchQueue.take();<a name="line.283"></a> +<span class="sourceLineNo">284</span> }<a name="line.284"></a> +<span class="sourceLineNo">285</span><a name="line.285"></a> +<span class="sourceLineNo">286</span> private long getEntrySize(Entry entry) {<a name="line.286"></a> +<span class="sourceLineNo">287</span> WALEdit edit = entry.getEdit();<a name="line.287"></a> +<span class="sourceLineNo">288</span> return edit.heapSize() + calculateTotalSizeOfStoreFiles(edit);<a name="line.288"></a> +<span class="sourceLineNo">289</span> }<a name="line.289"></a> +<span class="sourceLineNo">290</span><a name="line.290"></a> +<span class="sourceLineNo">291</span> private void updateBatchStats(WALEntryBatch batch, Entry entry, long entryPosition, long entrySize) {<a name="line.291"></a> +<span class="sourceLineNo">292</span> WALEdit edit = entry.getEdit();<a name="line.292"></a> +<span class="sourceLineNo">293</span> if (edit != null && !edit.isEmpty()) {<a name="line.293"></a> +<span class="sourceLineNo">294</span> batch.incrementHeapSize(entrySize);<a name="line.294"></a> +<span class="sourceLineNo">295</span> Pair<Integer, Integer> nbRowsAndHFiles = countDistinctRowKeysAndHFiles(edit);<a name="line.295"></a> +<span class="sourceLineNo">296</span> batch.incrementNbRowKeys(nbRowsAndHFiles.getFirst());<a name="line.296"></a> +<span class="sourceLineNo">297</span> batch.incrementNbHFiles(nbRowsAndHFiles.getSecond());<a name="line.297"></a> +<span class="sourceLineNo">298</span> }<a name="line.298"></a> +<span class="sourceLineNo">299</span> batch.lastWalPosition = entryPosition;<a name="line.299"></a> +<span class="sourceLineNo">300</span> }<a name="line.300"></a> +<span class="sourceLineNo">301</span><a name="line.301"></a> +<span class="sourceLineNo">302</span> /**<a name="line.302"></a> +<span class="sourceLineNo">303</span> * Count the number of different row keys in the given edit because of mini-batching. We assume<a name="line.303"></a> +<span class="sourceLineNo">304</span> * that there's at least one Cell in the WALEdit.<a name="line.304"></a> +<span class="sourceLineNo">305</span> * @param edit edit to count row keys from<a name="line.305"></a> +<span class="sourceLineNo">306</span> * @return number of different row keys and HFiles<a name="line.306"></a> +<span class="sourceLineNo">307</span> */<a name="line.307"></a> +<span class="sourceLineNo">308</span> private Pair<Integer, Integer> countDistinctRowKeysAndHFiles(WALEdit edit) {<a name="line.308"></a> +<span class="sourceLineNo">309</span> List<Cell> cells = edit.getCells();<a name="line.309"></a> +<span class="sourceLineNo">310</span> int distinctRowKeys = 1;<a name="line.310"></a> +<span class="sourceLineNo">311</span> int totalHFileEntries = 0;<a name="line.311"></a> +<span class="sourceLineNo">312</span> Cell lastCell = cells.get(0);<a name="line.312"></a> +<span class="sourceLineNo">313</span><a name="line.313"></a> +<span class="sourceLineNo">314</span> int totalCells = edit.size();<a name="line.314"></a> +<span class="sourceLineNo">315</span> for (int i = 0; i < totalCells; i++) {<a name="line.315"></a> +<span class="sourceLineNo">316</span> // Count HFiles to be replicated<a name="line.316"></a> +<span class="sourceLineNo">317</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.317"></a> +<span class="sourceLineNo">318</span> try {<a name="line.318"></a> +<span class="sourceLineNo">319</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.319"></a> +<span class="sourceLineNo">320</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.320"></a> +<span class="sourceLineNo">321</span> int totalStores = stores.size();<a name="line.321"></a> +<span class="sourceLineNo">322</span> for (int j = 0; j < totalStores; j++) {<a name="line.322"></a> +<span class="sourceLineNo">323</span> totalHFileEntries += stores.get(j).getStoreFileList().size();<a name="line.323"></a> +<span class="sourceLineNo">324</span> }<a name="line.324"></a> +<span class="sourceLineNo">325</span> } catch (IOException e) {<a name="line.325"></a> +<span class="sourceLineNo">326</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.326"></a> +<span class="sourceLineNo">327</span> + "Then its hfiles count will not be added into metric.");<a name="line.327"></a> +<span class="sourceLineNo">328</span> }<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> if (!CellUtil.matchingRows(cells.get(i), lastCell)) {<a name="line.331"></a> +<span class="sourceLineNo">332</span> distinctRowKeys++;<a name="line.332"></a> +<span class="sourceLineNo">333</span> }<a name="line.333"></a> +<span class="sourceLineNo">334</span> lastCell = cells.get(i);<a name="line.334"></a> +<span class="sourceLineNo">335</span> }<a name="line.335"></a> +<span class="sourceLineNo">336</span><a name="line.336"></a> +<span class="sourceLineNo">337</span> Pair<Integer, Integer> result = new Pair<>(distinctRowKeys, totalHFileEntries);<a name="line.337"></a> +<span class="sourceLineNo">338</span> return result;<a name="line.338"></a> <span class="sourceLineNo">339</span> }<a name="line.339"></a> <span class="sourceLineNo">340</span><a name="line.340"></a> <span class="sourceLineNo">341</span> /**<a name="line.341"></a> -<span class="sourceLineNo">342</span> * @param size delta size for grown buffer<a name="line.342"></a> -<span class="sourceLineNo">343</span> * @return true if we should clear buffer and push all<a name="line.343"></a> -<span class="sourceLineNo">344</span> */<a name="line.344"></a> -<span class="sourceLineNo">345</span> private boolean acquireBufferQuota(long size) {<a name="line.345"></a> -<span class="sourceLineNo">346</span> return totalBufferUsed.addAndGet(size) >= totalBufferQuota;<a name="line.346"></a> -<span class="sourceLineNo">347</span> }<a name="line.347"></a> -<span class="sourceLineNo">348</span><a name="line.348"></a> -<span class="sourceLineNo">349</span> /**<a name="line.349"></a> -<span class="sourceLineNo">350</span> * @return whether the reader thread is running<a name="line.350"></a> -<span class="sourceLineNo">351</span> */<a name="line.351"></a> -<span class="sourceLineNo">352</span> public boolean isReaderRunning() {<a name="line.352"></a> -<span class="sourceLineNo">353</span> return isReaderRunning && !isInterrupted();<a name="line.353"></a> -<span class="sourceLineNo">354</span> }<a name="line.354"></a> -<span class="sourceLineNo">355</span><a name="line.355"></a> -<span class="sourceLineNo">356</span> /**<a name="line.356"></a> -<span class="sourceLineNo">357</span> * @param readerRunning the readerRunning to set<a name="line.357"></a> -<span class="sourceLineNo">358</span> */<a name="line.358"></a> -<span class="sourceLineNo">359</span> public void setReaderRunning(boolean readerRunning) {<a name="line.359"></a> -<span class="sourceLineNo">360</span> this.isReaderRunning = readerRunning;<a name="line.360"></a> -<span class="sourceLineNo">361</span> }<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> * Holds a batch of WAL entries to replicate, along with some statistics<a name="line.364"></a> -<span class="sourceLineNo">365</span> *<a name="line.365"></a> -<span class="sourceLineNo">366</span> */<a name="line.366"></a> -<span class="sourceLineNo">367</span> static class WALEntryBatch {<a name="line.367"></a> -<span class="sourceLineNo">368</span> private List<Entry> walEntries;<a name="line.368"></a> -<span class="sourceLineNo">369</span> // last WAL that was read<a name="line.369"></a> -<span class="sourceLineNo">370</span> private Path lastWalPath;<a name="line.370"></a> -<span class="sourceLineNo">371</span> // position in WAL of last entry in this batch<a name="line.371"></a> -<span class="sourceLineNo">372</span> private long lastWalPosition = 0;<a name="line.372"></a> -<span class="sourceLineNo">373</span> // number of distinct row keys in this batch<a name="line.373"></a> -<span class="sourceLineNo">374</span> private int nbRowKeys = 0;<a name="line.374"></a> -<span class="sourceLineNo">375</span> // number of HFiles<a name="line.375"></a> -<span class="sourceLineNo">376</span> private int nbHFiles = 0;<a name="line.376"></a> -<span class="sourceLineNo">377</span> // heap size of data we need to replicate<a name="line.377"></a> -<span class="sourceLineNo">378</span> private long heapSize = 0;<a name="line.378"></a> -<span class="sourceLineNo">379</span> // save the last sequenceid for each region if the table has serial-replication scope<a name="line.379"></a> -<span class="sourceLineNo">380</span> private Map<String, Long> lastSeqIds = new HashMap<>();<a name="line.380"></a> -<span class="sourceLineNo">381</span><a name="line.381"></a> -<span class="sourceLineNo">382</span> /**<a name="line.382"></a> -<span class="sourceLineNo">383</span> * @param walEntries<a name="line.383"></a> -<span class="sourceLineNo">384</span> * @param lastWalPath Path of the WAL the last entry in this batch was read from<a name="line.384"></a> -<span class="sourceLineNo">385</span> * @param lastWalPosition Position in the WAL the last entry in this batch was read from<a name="line.385"></a> -<span class="sourceLineNo">386</span> */<a name="line.386"></a> -<span class="sourceLineNo">387</span> private WALEntryBatch(int maxNbEntries, Path lastWalPath) {<a name="line.387"></a> -<span class="sourceLineNo">388</span> this.walEntries = new ArrayList<>(maxNbEntries);<a name="line.388"></a> -<span class="sourceLineNo">389</span> this.lastWalPath = lastWalPath;<a name="line.389"></a> -<span class="sourceLineNo">390</span> }<a name="line.390"></a> -<span class="sourceLineNo">391</span><a name="line.391"></a> -<span class="sourceLineNo">392</span> public void addEntry(Entry entry) {<a name="line.392"></a> -<span class="sourceLineNo">393</span> walEntries.add(entry);<a name="line.393"></a> -<span class="sourceLineNo">394</span> }<a name="line.394"></a> -<span class="sourceLineNo">395</span><a name="line.395"></a> -<span class="sourceLineNo">396</span> /**<a name="line.396"></a> -<span class="sourceLineNo">397</span> * @return the WAL Entries.<a name="line.397"></a> -<span class="sourceLineNo">398</span> */<a name="line.398"></a> -<span class="sourceLineNo">399</span> public List<Entry> getWalEntries() {<a name="line.399"></a> -<span class="sourceLineNo">400</span> return walEntries;<a name="line.400"></a> -<span class="sourceLineNo">401</span> }<a name="line.401"></a> -<span class="sourceLineNo">402</span><a name="line.402"></a> -<span class="sourceLineNo">403</span> /**<a name="line.403"></a> -<span class="sourceLineNo">404</span> * @return the path of the last WAL that was read.<a name="line.404"></a> -<span class="sourceLineNo">405</span> */<a name="line.405"></a> -<span class="sourceLineNo">406</span> public Path getLastWalPath() {<a name="line.406"></a> -<span class="sourceLineNo">407</span> return lastWalPath;<a name="line.407"></a> -<span class="sourceLineNo">408</span> }<a name="line.408"></a> -<span class="sourceLineNo">409</span><a name="line.409"></a> -<span class="sourceLineNo">410</span> /**<a name="line.410"></a> -<span class="sourceLineNo">411</span> * @return the position in the last WAL that was read.<a name="line.411"></a> -<span class="sourceLineNo">412</span> */<a name="line.412"></a> -<span class="sourceLineNo">413</span> public long getLastWalPosition() {<a name="line.413"></a> -<span class="sourceLineNo">414</span> return lastWalPosition;<a name="line.414"></a> -<span class="sourceLineNo">415</span> }<a name="line.415"></a> -<span class="sourceLineNo">416</span><a name="line.416"></a> -<span class="sourceLineNo">417</span> public int getNbEntries() {<a name="line.417"></a> -<span class="sourceLineNo">418</span> return walEntries.size();<a name="line.418"></a> -<span class="sourceLineNo">419</span> }<a name="line.419"></a> -<span class="sourceLineNo">420</span><a name="line.420"></a> -<span class="sourceLineNo">421</span> /**<a name="line.421"></a> -<span class="sourceLineNo">422</span> * @return the number of distinct row keys in this batch<a name="line.422"></a> -<span class="sourceLineNo">423</span> */<a name="line.423"></a> -<span class="sourceLineNo">424</span> public int getNbRowKeys() {<a name="line.424"></a> -<span class="sourceLineNo">425</span> return nbRowKeys;<a name="line.425"></a> -<span class="sourceLineNo">426</span> }<a name="line.426"></a> -<span class="sourceLineNo">427</span><a name="line.427"></a> -<span class="sourceLineNo">428</span> /**<a name="line.428"></a> -<span class="sourceLineNo">429</span> * @return the number of HFiles in this batch<a name="line.429"></a> -<span class="sourceLineNo">430</span> */<a name="line.430"></a> -<span class="sourceLineNo">431</span> public int getNbHFiles() {<a name="line.431"></a> -<span class="sourceLineNo">432</span> return nbHFiles;<a name="line.432"></a> -<span class="sourceLineNo">433</span> }<a name="line.433"></a> -<span class="sourceLineNo">434</span><a name="line.434"></a> -<span class="sourceLineNo">435</span> /**<a name="line.435"></a> -<span class="sourceLineNo">436</span> * @return total number of operations in this batch<a name="line.436"></a> -<span class="sourceLineNo">437</span> */<a name="line.437"></a> -<span class="sourceLineNo">438</span> public int getNbOperations() {<a name="line.438"></a> -<span class="sourceLineNo">439</span> return getNbRowKeys() + getNbHFiles();<a name="line.439"></a> -<span class="sourceLineNo">440</span> }<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> * @return the heap size of this batch<a name="line.443"></a> -<span class="sourceLineNo">444</span> */<a name="line.444"></a> -<span class="sourceLineNo">445</span> public long getHeapSize() {<a name="line.445"></a> -<span class="sourceLineNo">446</span> return heapSize;<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> /**<a name="line.449"></a> -<span class="sourceLineNo">450</span> * @return the last sequenceid for each region if the table has serial-replication scope<a name="line.450"></a> -<span class="sourceLineNo">451</span> */<a name="line.451"></a> -<span class="sourceLineNo">452</span> public Map<String, Long> getLastSeqIds() {<a name="line.452"></a> -<span class="sourceLineNo">453</span> return lastSeqIds;<a name="line.453"></a> -<span class="sourceLineNo">454</span> }<a name="line.454"></a> -<span class="sourceLineNo">455</span><a name="line.455"></a> -<span class="sourceLineNo">456</span> private void incrementNbRowKeys(int increment) {<a name="line.456"></a> -<span class="sourceLineNo">457</span> nbRowKeys += increment;<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> private void incrementNbHFiles(int increment) {<a name="line.460"></a> -<span class="sourceLineNo">461</span> nbHFiles += increment;<a name="line.461"></a> -<span class="sourceLineNo">462</span> }<a name="line.462"></a> -<span class="sourceLineNo">463</span><a name="line.463"></a> -<span class="sourceLineNo">464</span> private void incrementHeapSize(long increment) {<a name="line.464"></a> -<span class="sourceLineNo">465</span> heapSize += increment;<a name="line.465"></a> -<span class="sourceLineNo">466</span> }<a name="line.466"></a> -<span class="sourceLineNo">467</span><a name="line.467"></a> -<span class="sourceLineNo">468</span> private void setLastPosition(String region, Long sequenceId) {<a name="line.468"></a> -<span class="sourceLineNo">469</span> getLastSeqIds().put(region, sequenceId);<a name="line.469"></a> +<span class="sourceLineNo">342</span> * Calculate the total size of all the store files<a name="line.342"></a> +<span class="sourceLineNo">343</span> * @param edit edit to count row keys from<a name="line.343"></a> +<span class="sourceLineNo">344</span> * @return the total size of the store files<a name="line.344"></a> +<span class="sourceLineNo">345</span> */<a name="line.345"></a> +<span class="sourceLineNo">346</span> private int calculateTotalSizeOfStoreFiles(WALEdit edit) {<a name="line.346"></a> +<span class="sourceLineNo">347</span> List<Cell> cells = edit.getCells();<a name="line.347"></a> +<span class="sourceLineNo">348</span> int totalStoreFilesSize = 0;<a name="line.348"></a> +<span class="sourceLineNo">349</span><a name="line.349"></a> +<span class="sourceLineNo">350</span> int totalCells = edit.size();<a name="line.350"></a> +<span class="sourceLineNo">351</span> for (int i = 0; i < totalCells; i++) {<a name="line.351"></a> +<span class="sourceLineNo">352</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.352"></a> +<span class="sourceLineNo">353</span> try {<a name="line.353"></a> +<span class="sourceLineNo">354</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.354"></a> +<span class="sourceLineNo">355</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.355"></a> +<span class="sourceLineNo">356</span> int totalStores = stores.size();<a name="line.356"></a> +<span class="sourceLineNo">357</span> for (int j = 0; j < totalStores; j++) {<a name="line.357"></a> +<span class="sourceLineNo">358</span> totalStoreFilesSize += stores.get(j).getStoreFileSizeBytes();<a name="line.358"></a> +<span class="sourceLineNo">359</span> }<a name="line.359"></a> +<span class="sourceLineNo">360</span> } catch (IOException e) {<a name="line.360"></a> +<span class="sourceLineNo">361</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.361"></a> +<span class="sourceLineNo">362</span> + "Size of HFiles part of cell will not be considered in replication "<a name="line.362"></a> +<span class="sourceLineNo">363</span> + "request size calculation.",<a name="line.363"></a> +<span class="sourceLineNo">364</span> e);<a name="line.364"></a> +<span class="sourceLineNo">365</span> }<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> return totalStoreFilesSize;<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> * @param size delta size for grown buffer<a name="line.372"></a> +<span class="sourceLineNo">373</span> * @return true if we should clear buffer and push all<a name="line.373"></a> +<span class="sourceLineNo">374</span> */<a name="line.374"></a> +<span class="sourceLineNo">375</span> private boolean acquireBufferQuota(long size) {<a name="line.375"></a> +<span class="sourceLineNo">376</span> return totalBufferUsed.addAndGet(size) >= totalBufferQuota;<a name="line.376"></a> +<span class="sourceLineNo">377</span> }<a name="line.377"></a> +<span class="sourceLineNo">378</span><a name="line.378"></a> +<span class="sourceLineNo">379</span> /**<a name="line.379"></a> +<span class="sourceLineNo">380</span> * @return whether the reader thread is running<a name="line.380"></a> +<span class="sourceLineNo">381</span> */<a name="line.381"></a> +<span class="sourceLineNo">382</span> public boolean isReaderRunning() {<a name="line.382"></a> +<span class="sourceLineNo">383</span> return isReaderRunning && !isInterrupted();<a name="line.383"></a> +<span class="sourceLineNo">384</span> }<a name="line.384"></a> +<span class="sourceLineNo">385</span><a name="line.385"></a> +<span class="sourceLineNo">386</span> /**<a name="line.386"></a> +<span class="sourceLineNo">387</span> * @param readerRunning the readerRunning to set<a name="line.387"></a> +<span class="sourceLineNo">388</span> */<a name="line.388"></a> +<span class="sourceLineNo">389</span> public void setReaderRunning(boolean readerRunning) {<a name="line.389"></a> +<span class="sourceLineNo">390</span> this.isReaderRunning = readerRunning;<a name="line.390"></a> +<span class="sourceLineNo">391</span> }<a name="line.391"></a> +<span class="sourceLineNo">392</span><a name="line.392"></a> +<span class="sourceLineNo">393</span> /**<a name="line.393"></a> +<span class="sourceLineNo">394</span> * Holds a batch of WAL entries to replicate, along with some statistics<a name="line.394"></a> +<span class="sourceLineNo">395</span> *<a name="line.395"></a> +<span class="sourceLineNo">396</span> */<a name="line.396"></a> +<span class="sourceLineNo">397</span> static class WALEntryBatch {<a name="line.397"></a> +<span class="sourceLineNo">398</span> private List<Entry> walEntries;<a name="line.398"></a> +<span class="sourceLineNo">399</span> // last WAL that was read<a name="line.399"></a> +<span class="sourceLineNo">400</span> private Path lastWalPath;<a name="line.400"></a> +<span class="sourceLineNo">401</span> // position in WAL of last entry in this batch<a name="line.401"></a> +<span class="sourceLineNo">402</span> private long lastWalPosition = 0;<a name="line.402"></a> +<span class="sourceLineNo">403</span> // number of distinct row keys in this batch<a name="line.403"></a> +<span class="sourceLineNo">404</span> private int nbRowKeys = 0;<a name="line.404"></a> +<span class="sourceLineNo">405</span> // number of HFiles<a name="line.405"></a> +<span class="sourceLineNo">406</span> private int nbHFiles = 0;<a name="line.406"></a> +<span class="sourceLineNo">407</span> // heap size of data we need to replicate<a name="line.407"></a> +<span class="sourceLineNo">408</span> private long heapSize = 0;<a name="line.408"></a> +<span class="sourceLineNo">409</span> // save the last sequenceid for each region if the table has serial-replication scope<a name="line.409"></a> +<span class="sourceLineNo">410</span> private Map<String, Long> lastSeqIds = new HashMap<>();<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> * @param walEntries<a name="line.413"></a> +<span class="sourceLineNo">414</span> * @param lastWalPath Path of the WAL the last entry in this batch was read from<a name="line.414"></a> +<span class="sourceLineNo">415</span> * @param lastWalPosition Position in the WAL the last entry in this batch was read from<a name="line.415"></a> +<span class="sourceLineNo">416</span> */<a name="line.416"></a> +<span class="sourceLineNo">417</span> private WALEntryBatch(int maxNbEntries, Path lastWalPath) {<a name="line.417"></a> +<span class="sourceLineNo">418</span> this.walEntries = new ArrayList<>(maxNbEntries);<a name="line.418"></a> +<span class="sourceLineNo">419</span> this.lastWalPath = lastWalPath;<a name="line.419"></a> +<span class="sourceLineNo">420</span> }<a name="line.420"></a> +<span class="sourceLineNo">421</span><a name="line.421"></a> +<span class="sourceLineNo">422</span> public void addEntry(Entry entry) {<a name="line.422"></a> +<span class="sourceLineNo">423</span> walEntries.add(entry);<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> * @return the WAL Entries.<a name="line.427"></a> +<span class="sourceLineNo">428</span> */<a name="line.428"></a> +<span class="sourceLineNo">429</span> public List<Entry> getWalEntries() {<a name="line.429"></a> +<span class="sourceLineNo">430</span> return walEntries;<a name="line.430"></a> +<span class="sourceLineNo">431</span> }<a name="line.431"></a> +<span class="sourceLineNo">432</span><a name="line.432"></a> +<span class="sourceLineNo">433</span> /**<a name="line.433"></a> +<span class="sourceLineNo">434</span> * @return the path of the last WAL that was read.<a name="line.434"></a> +<span class="sourceLineNo">435</span> */<a name="line.435"></a> +<span class="sourceLineNo">436</span> public Path getLastWalPath() {<a name="line.436"></a> +<span class="sourceLineNo">437</span> return lastWalPath;<a name="line.437"></a> +<span class="sourceLineNo">438</span> }<a name="line.438"></a> +<span class="sourceLineNo">439</span><a name="line.439"></a> +<span class="sourceLineNo">440</span> /**<a name="line.440"></a> +<span class="sourceLineNo">441</span> * @return the position in the last WAL that was read.<a name="line.441"></a> +<span class="sourceLineNo">442</span> */<a name="line.442"></a> +<span class="sourceLineNo">443</span> public long getLastWalPosition() {<a name="line.443"></a> +<span class="sourceLineNo">444</span> return lastWalPosition;<a name="line.444"></a> +<span class="sourceLineNo">445</span> }<a name="line.445"></a> +<span class="sourceLineNo">446</span><a name="line.446"></a> +<span class="sourceLineNo">447</span> public int getNbEntries() {<a name="line.447"></a> +<span class="sourceLineNo">448</span> return walEntries.size();<a name="line.448"></a> +<span class="sourceLineNo">449</span> }<a name="line.449"></a> +<span class="sourceLineNo">450</span><a name="line.450"></a> +<span class="sourceLineNo">451</span> /**<a name="line.451"></a> +<span class="sourceLineNo">452</span> * @return the number of distinct row keys in this batch<a name="line.452"></a> +<span class="sourceLineNo">453</span> */<a name="line.453"></a> +<span class="sourceLineNo">454</span> public int getNbRowKeys() {<a name="line.454"></a> +<span class="sourceLineNo">455</span> return nbRowKeys;<a name="line.455"></a> +<span class="sourceLineNo">456</span> }<a name="line.456"></a> +<span class="sourceLineNo">457</span><a name="line.457"></a> +<span class="sourceLineNo">458</span> /**<a name="line.458"></a> +<span class="sourceLineNo">459</span> * @return the number of HFiles in
<TRUNCATED>
