http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f1b07859/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReader.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReader.html b/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReader.html index 02aead0..be437fe 100644 --- a/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReader.html +++ b/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/ReplicationSourceWALReader.html @@ -40,383 +40,336 @@ <span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.32"></a> <span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.HConstants;<a name="line.33"></a> <span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.replication.WALEntryFilter;<a name="line.34"></a> -<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.35"></a> -<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.36"></a> -<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.util.Threads;<a name="line.37"></a> -<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<a name="line.38"></a> -<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.wal.WALEdit;<a name="line.39"></a> -<span class="sourceLineNo">040</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.40"></a> -<span class="sourceLineNo">041</span>import org.apache.yetus.audience.InterfaceStability;<a name="line.41"></a> -<span class="sourceLineNo">042</span>import org.slf4j.Logger;<a name="line.42"></a> -<span class="sourceLineNo">043</span>import org.slf4j.LoggerFactory;<a name="line.43"></a> -<span class="sourceLineNo">044</span><a name="line.44"></a> -<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.BulkLoadDescriptor;<a name="line.45"></a> -<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.StoreDescriptor;<a name="line.46"></a> -<span class="sourceLineNo">047</span><a name="line.47"></a> -<span class="sourceLineNo">048</span>/**<a name="line.48"></a> -<span class="sourceLineNo">049</span> * Reads and filters WAL entries, groups the filtered entries into batches, and puts the batches<a name="line.49"></a> -<span class="sourceLineNo">050</span> * onto a queue<a name="line.50"></a> -<span class="sourceLineNo">051</span> */<a name="line.51"></a> -<span class="sourceLineNo">052</span>@InterfaceAudience.Private<a name="line.52"></a> -<span class="sourceLineNo">053</span>@InterfaceStability.Evolving<a name="line.53"></a> -<span class="sourceLineNo">054</span>public class ReplicationSourceWALReader extends Thread {<a name="line.54"></a> -<span class="sourceLineNo">055</span> private static final Logger LOG = LoggerFactory.getLogger(ReplicationSourceWALReader.class);<a name="line.55"></a> -<span class="sourceLineNo">056</span><a name="line.56"></a> -<span class="sourceLineNo">057</span> private final PriorityBlockingQueue<Path> logQueue;<a name="line.57"></a> -<span class="sourceLineNo">058</span> private final FileSystem fs;<a name="line.58"></a> -<span class="sourceLineNo">059</span> private final Configuration conf;<a name="line.59"></a> -<span class="sourceLineNo">060</span> private final WALEntryFilter filter;<a name="line.60"></a> -<span class="sourceLineNo">061</span> private final ReplicationSource source;<a name="line.61"></a> -<span class="sourceLineNo">062</span><a name="line.62"></a> -<span class="sourceLineNo">063</span> protected final BlockingQueue<WALEntryBatch> entryBatchQueue;<a name="line.63"></a> -<span class="sourceLineNo">064</span> // max (heap) size of each batch - multiply by number of batches in queue to get total<a name="line.64"></a> -<span class="sourceLineNo">065</span> private final long replicationBatchSizeCapacity;<a name="line.65"></a> -<span class="sourceLineNo">066</span> // max count of each batch - multiply by number of batches in queue to get total<a name="line.66"></a> -<span class="sourceLineNo">067</span> protected final int replicationBatchCountCapacity;<a name="line.67"></a> -<span class="sourceLineNo">068</span> // position in the WAL to start reading at<a name="line.68"></a> -<span class="sourceLineNo">069</span> private long currentPosition;<a name="line.69"></a> -<span class="sourceLineNo">070</span> private final long sleepForRetries;<a name="line.70"></a> -<span class="sourceLineNo">071</span> private final int maxRetriesMultiplier;<a name="line.71"></a> -<span class="sourceLineNo">072</span> private final boolean eofAutoRecovery;<a name="line.72"></a> -<span class="sourceLineNo">073</span><a name="line.73"></a> -<span class="sourceLineNo">074</span> // used to store the first cell in an entry before filtering. This is because that if serial<a name="line.74"></a> -<span class="sourceLineNo">075</span> // replication is enabled, we may find out that an entry can not be pushed after filtering. And<a name="line.75"></a> -<span class="sourceLineNo">076</span> // when we try the next time, the cells maybe null since the entry has already been filtered,<a name="line.76"></a> -<span class="sourceLineNo">077</span> // especially for region event wal entries. And this can also used to determine whether we can<a name="line.77"></a> -<span class="sourceLineNo">078</span> // skip filtering.<a name="line.78"></a> -<span class="sourceLineNo">079</span> private Cell firstCellInEntryBeforeFiltering;<a name="line.79"></a> -<span class="sourceLineNo">080</span><a name="line.80"></a> -<span class="sourceLineNo">081</span> //Indicates whether this particular worker is running<a name="line.81"></a> -<span class="sourceLineNo">082</span> private boolean isReaderRunning = true;<a name="line.82"></a> -<span class="sourceLineNo">083</span><a name="line.83"></a> -<span class="sourceLineNo">084</span> private AtomicLong totalBufferUsed;<a name="line.84"></a> -<span class="sourceLineNo">085</span> private long totalBufferQuota;<a name="line.85"></a> -<span class="sourceLineNo">086</span><a name="line.86"></a> -<span class="sourceLineNo">087</span> private final SerialReplicationChecker serialReplicationChecker;<a name="line.87"></a> -<span class="sourceLineNo">088</span><a name="line.88"></a> -<span class="sourceLineNo">089</span> /**<a name="line.89"></a> -<span class="sourceLineNo">090</span> * Creates a reader worker for a given WAL queue. Reads WAL entries off a given queue, batches the<a name="line.90"></a> -<span class="sourceLineNo">091</span> * entries, and puts them on a batch queue.<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 logQueue The WAL queue to read off of<a name="line.94"></a> -<span class="sourceLineNo">095</span> * @param startPosition position in the first WAL to start reading from<a name="line.95"></a> -<span class="sourceLineNo">096</span> * @param filter The filter to use while reading<a name="line.96"></a> -<span class="sourceLineNo">097</span> * @param source replication source<a name="line.97"></a> -<span class="sourceLineNo">098</span> */<a name="line.98"></a> -<span class="sourceLineNo">099</span> public ReplicationSourceWALReader(FileSystem fs, Configuration conf,<a name="line.99"></a> -<span class="sourceLineNo">100</span> PriorityBlockingQueue<Path> logQueue, long startPosition, WALEntryFilter filter,<a name="line.100"></a> -<span class="sourceLineNo">101</span> ReplicationSource source) {<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.source = source;<a name="line.107"></a> -<span class="sourceLineNo">108</span> this.replicationBatchSizeCapacity =<a name="line.108"></a> -<span class="sourceLineNo">109</span> this.conf.getLong("replication.source.size.capacity", 1024 * 1024 * 64);<a name="line.109"></a> -<span class="sourceLineNo">110</span> this.replicationBatchCountCapacity = this.conf.getInt("replication.source.nb.capacity", 25000);<a name="line.110"></a> -<span class="sourceLineNo">111</span> // memory used will be batchSizeCapacity * (nb.batches + 1)<a name="line.111"></a> -<span class="sourceLineNo">112</span> // the +1 is for the current thread reading before placing onto the queue<a name="line.112"></a> -<span class="sourceLineNo">113</span> int batchCount = conf.getInt("replication.source.nb.batches", 1);<a name="line.113"></a> -<span class="sourceLineNo">114</span> this.totalBufferUsed = source.getSourceManager().getTotalBufferUsed();<a name="line.114"></a> -<span class="sourceLineNo">115</span> this.totalBufferQuota = conf.getLong(HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_KEY,<a name="line.115"></a> -<span class="sourceLineNo">116</span> HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_DFAULT);<a name="line.116"></a> -<span class="sourceLineNo">117</span> this.sleepForRetries =<a name="line.117"></a> -<span class="sourceLineNo">118</span> this.conf.getLong("replication.source.sleepforretries", 1000); // 1 second<a name="line.118"></a> -<span class="sourceLineNo">119</span> this.maxRetriesMultiplier =<a name="line.119"></a> -<span class="sourceLineNo">120</span> this.conf.getInt("replication.source.maxretriesmultiplier", 300); // 5 minutes @ 1 sec per<a name="line.120"></a> -<span class="sourceLineNo">121</span> this.eofAutoRecovery = conf.getBoolean("replication.source.eof.autorecovery", false);<a name="line.121"></a> -<span class="sourceLineNo">122</span> this.entryBatchQueue = new LinkedBlockingQueue<>(batchCount);<a name="line.122"></a> -<span class="sourceLineNo">123</span> this.serialReplicationChecker = new SerialReplicationChecker(conf, source);<a name="line.123"></a> -<span class="sourceLineNo">124</span> LOG.info("peerClusterZnode=" + source.getQueueId()<a name="line.124"></a> -<span class="sourceLineNo">125</span> + ", ReplicationSourceWALReaderThread : " + source.getPeerId()<a name="line.125"></a> -<span class="sourceLineNo">126</span> + " inited, replicationBatchSizeCapacity=" + replicationBatchSizeCapacity<a name="line.126"></a> -<span class="sourceLineNo">127</span> + ", replicationBatchCountCapacity=" + replicationBatchCountCapacity<a name="line.127"></a> -<span class="sourceLineNo">128</span> + ", replicationBatchQueueCapacity=" + batchCount);<a name="line.128"></a> -<span class="sourceLineNo">129</span> }<a name="line.129"></a> -<span class="sourceLineNo">130</span><a name="line.130"></a> -<span class="sourceLineNo">131</span> @Override<a name="line.131"></a> -<span class="sourceLineNo">132</span> public void run() {<a name="line.132"></a> -<span class="sourceLineNo">133</span> int sleepMultiplier = 1;<a name="line.133"></a> -<span class="sourceLineNo">134</span> while (isReaderRunning()) { // we only loop back here if something fatal happened to our stream<a name="line.134"></a> -<span class="sourceLineNo">135</span> try (WALEntryStream entryStream =<a name="line.135"></a> -<span class="sourceLineNo">136</span> new WALEntryStream(logQueue, fs, conf, currentPosition,<a name="line.136"></a> -<span class="sourceLineNo">137</span> source.getWALFileLengthProvider(), source.getServerWALsBelongTo(),<a name="line.137"></a> -<span class="sourceLineNo">138</span> source.getSourceMetrics())) {<a name="line.138"></a> -<span class="sourceLineNo">139</span> while (isReaderRunning()) { // loop here to keep reusing stream while we can<a name="line.139"></a> -<span class="sourceLineNo">140</span> if (!checkQuota()) {<a name="line.140"></a> -<span class="sourceLineNo">141</span> continue;<a name="line.141"></a> -<span class="sourceLineNo">142</span> }<a name="line.142"></a> -<span class="sourceLineNo">143</span> WALEntryBatch batch = readWALEntries(entryStream);<a name="line.143"></a> -<span class="sourceLineNo">144</span> if (batch != null) {<a name="line.144"></a> -<span class="sourceLineNo">145</span> // need to propagate the batch even it has no entries since it may carry the last<a name="line.145"></a> -<span class="sourceLineNo">146</span> // sequence id information for serial replication.<a name="line.146"></a> -<span class="sourceLineNo">147</span> LOG.trace("Read {} WAL entries eligible for replication", batch.getNbEntries());<a name="line.147"></a> -<span class="sourceLineNo">148</span> entryBatchQueue.put(batch);<a name="line.148"></a> -<span class="sourceLineNo">149</span> sleepMultiplier = 1;<a name="line.149"></a> -<span class="sourceLineNo">150</span> } else { // got no entries and didn't advance position in WAL<a name="line.150"></a> -<span class="sourceLineNo">151</span> handleEmptyWALEntryBatch(entryStream.getCurrentPath());<a name="line.151"></a> -<span class="sourceLineNo">152</span> }<a name="line.152"></a> -<span class="sourceLineNo">153</span> currentPosition = entryStream.getPosition();<a name="line.153"></a> -<span class="sourceLineNo">154</span> entryStream.reset(); // reuse stream<a name="line.154"></a> -<span class="sourceLineNo">155</span> }<a name="line.155"></a> -<span class="sourceLineNo">156</span> } catch (IOException e) { // stream related<a name="line.156"></a> -<span class="sourceLineNo">157</span> if (sleepMultiplier < maxRetriesMultiplier) {<a name="line.157"></a> -<span class="sourceLineNo">158</span> LOG.debug("Failed to read stream of replication entries: " + e);<a name="line.158"></a> -<span class="sourceLineNo">159</span> sleepMultiplier++;<a name="line.159"></a> -<span class="sourceLineNo">160</span> } else {<a name="line.160"></a> -<span class="sourceLineNo">161</span> LOG.error("Failed to read stream of replication entries", e);<a name="line.161"></a> -<span class="sourceLineNo">162</span> handleEofException(e);<a name="line.162"></a> -<span class="sourceLineNo">163</span> }<a name="line.163"></a> -<span class="sourceLineNo">164</span> Threads.sleep(sleepForRetries * sleepMultiplier);<a name="line.164"></a> -<span class="sourceLineNo">165</span> } catch (InterruptedException e) {<a name="line.165"></a> -<span class="sourceLineNo">166</span> LOG.trace("Interrupted while sleeping between WAL reads");<a name="line.166"></a> -<span class="sourceLineNo">167</span> Thread.currentThread().interrupt();<a name="line.167"></a> -<span class="sourceLineNo">168</span> }<a name="line.168"></a> -<span class="sourceLineNo">169</span> }<a name="line.169"></a> -<span class="sourceLineNo">170</span> }<a name="line.170"></a> -<span class="sourceLineNo">171</span><a name="line.171"></a> -<span class="sourceLineNo">172</span> private void removeEntryFromStream(WALEntryStream entryStream, WALEntryBatch batch)<a name="line.172"></a> -<span class="sourceLineNo">173</span> throws IOException {<a name="line.173"></a> -<span class="sourceLineNo">174</span> entryStream.next();<a name="line.174"></a> -<span class="sourceLineNo">175</span> firstCellInEntryBeforeFiltering = null;<a name="line.175"></a> -<span class="sourceLineNo">176</span> batch.setLastWalPosition(entryStream.getPosition());<a name="line.176"></a> -<span class="sourceLineNo">177</span> }<a name="line.177"></a> -<span class="sourceLineNo">178</span><a name="line.178"></a> -<span class="sourceLineNo">179</span> private WALEntryBatch readWALEntries(WALEntryStream entryStream)<a name="line.179"></a> -<span class="sourceLineNo">180</span> throws IOException, InterruptedException {<a name="line.180"></a> -<span class="sourceLineNo">181</span> if (!entryStream.hasNext()) {<a name="line.181"></a> -<span class="sourceLineNo">182</span> return null;<a name="line.182"></a> -<span class="sourceLineNo">183</span> }<a name="line.183"></a> -<span class="sourceLineNo">184</span> long positionBefore = entryStream.getPosition();<a name="line.184"></a> -<span class="sourceLineNo">185</span> WALEntryBatch batch =<a name="line.185"></a> -<span class="sourceLineNo">186</span> new WALEntryBatch(replicationBatchCountCapacity, entryStream.getCurrentPath());<a name="line.186"></a> -<span class="sourceLineNo">187</span> do {<a name="line.187"></a> -<span class="sourceLineNo">188</span> Entry entry = entryStream.peek();<a name="line.188"></a> -<span class="sourceLineNo">189</span> boolean isSerial = source.isSerial();<a name="line.189"></a> -<span class="sourceLineNo">190</span> boolean doFiltering = true;<a name="line.190"></a> -<span class="sourceLineNo">191</span> if (isSerial) {<a name="line.191"></a> -<span class="sourceLineNo">192</span> if (firstCellInEntryBeforeFiltering == null) {<a name="line.192"></a> -<span class="sourceLineNo">193</span> assert !entry.getEdit().isEmpty() : "should not write empty edits";<a name="line.193"></a> -<span class="sourceLineNo">194</span> // Used to locate the region record in meta table. In WAL we only have the table name and<a name="line.194"></a> -<span class="sourceLineNo">195</span> // encoded region name which can not be mapping to region name without scanning all the<a name="line.195"></a> -<span class="sourceLineNo">196</span> // records for a table, so we need a start key, just like what we have done at client side<a name="line.196"></a> -<span class="sourceLineNo">197</span> // when locating a region. For the markers, we will use the start key of the region as the<a name="line.197"></a> -<span class="sourceLineNo">198</span> // row key for the edit. And we need to do this before filtering since all the cells may<a name="line.198"></a> -<span class="sourceLineNo">199</span> // be filtered out, especially that for the markers.<a name="line.199"></a> -<span class="sourceLineNo">200</span> firstCellInEntryBeforeFiltering = entry.getEdit().getCells().get(0);<a name="line.200"></a> -<span class="sourceLineNo">201</span> } else {<a name="line.201"></a> -<span class="sourceLineNo">202</span> // if this is not null then we know that the entry has already been filtered.<a name="line.202"></a> -<span class="sourceLineNo">203</span> doFiltering = false;<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> if (doFiltering) {<a name="line.207"></a> -<span class="sourceLineNo">208</span> entry = filterEntry(entry);<a name="line.208"></a> -<span class="sourceLineNo">209</span> }<a name="line.209"></a> -<span class="sourceLineNo">210</span> if (entry != null) {<a name="line.210"></a> -<span class="sourceLineNo">211</span> if (isSerial) {<a name="line.211"></a> -<span class="sourceLineNo">212</span> if (!serialReplicationChecker.canPush(entry, firstCellInEntryBeforeFiltering)) {<a name="line.212"></a> -<span class="sourceLineNo">213</span> if (batch.getLastWalPosition() > positionBefore) {<a name="line.213"></a> -<span class="sourceLineNo">214</span> // we have something that can push, break<a name="line.214"></a> -<span class="sourceLineNo">215</span> break;<a name="line.215"></a> -<span class="sourceLineNo">216</span> } else {<a name="line.216"></a> -<span class="sourceLineNo">217</span> serialReplicationChecker.waitUntilCanPush(entry, firstCellInEntryBeforeFiltering);<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> // arrive here means we can push the entry, record the last sequence id<a name="line.220"></a> -<span class="sourceLineNo">221</span> batch.setLastSeqId(Bytes.toString(entry.getKey().getEncodedRegionName()),<a name="line.221"></a> -<span class="sourceLineNo">222</span> entry.getKey().getSequenceId());<a name="line.222"></a> -<span class="sourceLineNo">223</span> }<a name="line.223"></a> -<span class="sourceLineNo">224</span> // actually remove the entry.<a name="line.224"></a> -<span class="sourceLineNo">225</span> removeEntryFromStream(entryStream, batch);<a name="line.225"></a> -<span class="sourceLineNo">226</span> WALEdit edit = entry.getEdit();<a name="line.226"></a> -<span class="sourceLineNo">227</span> if (edit != null && !edit.isEmpty()) {<a name="line.227"></a> -<span class="sourceLineNo">228</span> long entrySize = getEntrySize(entry);<a name="line.228"></a> -<span class="sourceLineNo">229</span> batch.addEntry(entry);<a name="line.229"></a> -<span class="sourceLineNo">230</span> updateBatchStats(batch, entry, entrySize);<a name="line.230"></a> -<span class="sourceLineNo">231</span> boolean totalBufferTooLarge = acquireBufferQuota(entrySize);<a name="line.231"></a> -<span class="sourceLineNo">232</span> // Stop if too many entries or too big<a name="line.232"></a> -<span class="sourceLineNo">233</span> if (totalBufferTooLarge || batch.getHeapSize() >= replicationBatchSizeCapacity ||<a name="line.233"></a> -<span class="sourceLineNo">234</span> batch.getNbEntries() >= replicationBatchCountCapacity) {<a name="line.234"></a> -<span class="sourceLineNo">235</span> break;<a name="line.235"></a> -<span class="sourceLineNo">236</span> }<a name="line.236"></a> -<span class="sourceLineNo">237</span> }<a name="line.237"></a> -<span class="sourceLineNo">238</span> } else {<a name="line.238"></a> -<span class="sourceLineNo">239</span> // actually remove the entry.<a name="line.239"></a> -<span class="sourceLineNo">240</span> removeEntryFromStream(entryStream, batch);<a name="line.240"></a> -<span class="sourceLineNo">241</span> }<a name="line.241"></a> -<span class="sourceLineNo">242</span> } while (entryStream.hasNext());<a name="line.242"></a> -<span class="sourceLineNo">243</span> return batch;<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> protected void handleEmptyWALEntryBatch(Path currentPath) throws InterruptedException {<a name="line.246"></a> -<span class="sourceLineNo">247</span> LOG.trace("Didn't read any new entries from WAL");<a name="line.247"></a> -<span class="sourceLineNo">248</span> Thread.sleep(sleepForRetries);<a name="line.248"></a> -<span class="sourceLineNo">249</span> }<a name="line.249"></a> -<span class="sourceLineNo">250</span><a name="line.250"></a> -<span class="sourceLineNo">251</span> // if we get an EOF due to a zero-length log, and there are other logs in queue<a name="line.251"></a> -<span class="sourceLineNo">252</span> // (highly likely we've closed the current log), we've hit the max retries, and autorecovery is<a name="line.252"></a> -<span class="sourceLineNo">253</span> // enabled, then dump the log<a name="line.253"></a> -<span class="sourceLineNo">254</span> private void handleEofException(IOException e) {<a name="line.254"></a> -<span class="sourceLineNo">255</span> if ((e instanceof EOFException || e.getCause() instanceof EOFException) &&<a name="line.255"></a> -<span class="sourceLineNo">256</span> logQueue.size() > 1 && this.eofAutoRecovery) {<a name="line.256"></a> -<span class="sourceLineNo">257</span> try {<a name="line.257"></a> -<span class="sourceLineNo">258</span> if (fs.getFileStatus(logQueue.peek()).getLen() == 0) {<a name="line.258"></a> -<span class="sourceLineNo">259</span> LOG.warn("Forcing removal of 0 length log in queue: " + logQueue.peek());<a name="line.259"></a> -<span class="sourceLineNo">260</span> logQueue.remove();<a name="line.260"></a> -<span class="sourceLineNo">261</span> currentPosition = 0;<a name="line.261"></a> -<span class="sourceLineNo">262</span> }<a name="line.262"></a> -<span class="sourceLineNo">263</span> } catch (IOException ioe) {<a name="line.263"></a> -<span class="sourceLineNo">264</span> LOG.warn("Couldn't get file length information about log " + logQueue.peek());<a name="line.264"></a> -<span class="sourceLineNo">265</span> }<a name="line.265"></a> -<span class="sourceLineNo">266</span> }<a name="line.266"></a> -<span class="sourceLineNo">267</span> }<a name="line.267"></a> -<span class="sourceLineNo">268</span><a name="line.268"></a> -<span class="sourceLineNo">269</span> public Path getCurrentPath() {<a name="line.269"></a> -<span class="sourceLineNo">270</span> // if we've read some WAL entries, get the Path we read from<a name="line.270"></a> -<span class="sourceLineNo">271</span> WALEntryBatch batchQueueHead = entryBatchQueue.peek();<a name="line.271"></a> -<span class="sourceLineNo">272</span> if (batchQueueHead != null) {<a name="line.272"></a> -<span class="sourceLineNo">273</span> return batchQueueHead.getLastWalPath();<a name="line.273"></a> -<span class="sourceLineNo">274</span> }<a name="line.274"></a> -<span class="sourceLineNo">275</span> // otherwise, we must be currently reading from the head of the log queue<a name="line.275"></a> -<span class="sourceLineNo">276</span> return logQueue.peek();<a name="line.276"></a> -<span class="sourceLineNo">277</span> }<a name="line.277"></a> -<span class="sourceLineNo">278</span><a name="line.278"></a> -<span class="sourceLineNo">279</span> //returns false if we've already exceeded the global quota<a name="line.279"></a> -<span class="sourceLineNo">280</span> private boolean checkQuota() {<a name="line.280"></a> -<span class="sourceLineNo">281</span> // try not to go over total quota<a name="line.281"></a> -<span class="sourceLineNo">282</span> if (totalBufferUsed.get() > totalBufferQuota) {<a name="line.282"></a> -<span class="sourceLineNo">283</span> Threads.sleep(sleepForRetries);<a name="line.283"></a> -<span class="sourceLineNo">284</span> return false;<a name="line.284"></a> -<span class="sourceLineNo">285</span> }<a name="line.285"></a> -<span class="sourceLineNo">286</span> return true;<a name="line.286"></a> -<span class="sourceLineNo">287</span> }<a name="line.287"></a> -<span class="sourceLineNo">288</span><a name="line.288"></a> -<span class="sourceLineNo">289</span> private Entry filterEntry(Entry entry) {<a name="line.289"></a> -<span class="sourceLineNo">290</span> Entry filtered = filter.filter(entry);<a name="line.290"></a> -<span class="sourceLineNo">291</span> if (entry != null && filtered == null) {<a name="line.291"></a> -<span class="sourceLineNo">292</span> source.getSourceMetrics().incrLogEditsFiltered();<a name="line.292"></a> -<span class="sourceLineNo">293</span> }<a name="line.293"></a> -<span class="sourceLineNo">294</span> return filtered;<a name="line.294"></a> -<span class="sourceLineNo">295</span> }<a name="line.295"></a> -<span class="sourceLineNo">296</span><a name="line.296"></a> -<span class="sourceLineNo">297</span> /**<a name="line.297"></a> -<span class="sourceLineNo">298</span> * Retrieves the next batch of WAL entries from the queue, waiting up to the specified time for a<a name="line.298"></a> -<span class="sourceLineNo">299</span> * batch to become available<a name="line.299"></a> -<span class="sourceLineNo">300</span> * @return A batch of entries, along with the position in the log after reading the batch<a name="line.300"></a> -<span class="sourceLineNo">301</span> * @throws InterruptedException if interrupted while waiting<a name="line.301"></a> -<span class="sourceLineNo">302</span> */<a name="line.302"></a> -<span class="sourceLineNo">303</span> public WALEntryBatch take() throws InterruptedException {<a name="line.303"></a> -<span class="sourceLineNo">304</span> return entryBatchQueue.take();<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> private long getEntrySize(Entry entry) {<a name="line.307"></a> -<span class="sourceLineNo">308</span> WALEdit edit = entry.getEdit();<a name="line.308"></a> -<span class="sourceLineNo">309</span> return edit.heapSize() + calculateTotalSizeOfStoreFiles(edit);<a name="line.309"></a> +<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.35"></a> +<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.util.Threads;<a name="line.36"></a> +<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<a name="line.37"></a> +<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.wal.WALEdit;<a name="line.38"></a> +<span class="sourceLineNo">039</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.39"></a> +<span class="sourceLineNo">040</span>import org.apache.yetus.audience.InterfaceStability;<a name="line.40"></a> +<span class="sourceLineNo">041</span>import org.slf4j.Logger;<a name="line.41"></a> +<span class="sourceLineNo">042</span>import org.slf4j.LoggerFactory;<a name="line.42"></a> +<span class="sourceLineNo">043</span><a name="line.43"></a> +<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.BulkLoadDescriptor;<a name="line.44"></a> +<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.WALProtos.StoreDescriptor;<a name="line.45"></a> +<span class="sourceLineNo">046</span><a name="line.46"></a> +<span class="sourceLineNo">047</span>/**<a name="line.47"></a> +<span class="sourceLineNo">048</span> * Reads and filters WAL entries, groups the filtered entries into batches, and puts the batches<a name="line.48"></a> +<span class="sourceLineNo">049</span> * onto a queue<a name="line.49"></a> +<span class="sourceLineNo">050</span> */<a name="line.50"></a> +<span class="sourceLineNo">051</span>@InterfaceAudience.Private<a name="line.51"></a> +<span class="sourceLineNo">052</span>@InterfaceStability.Evolving<a name="line.52"></a> +<span class="sourceLineNo">053</span>class ReplicationSourceWALReader extends Thread {<a name="line.53"></a> +<span class="sourceLineNo">054</span> private static final Logger LOG = LoggerFactory.getLogger(ReplicationSourceWALReader.class);<a name="line.54"></a> +<span class="sourceLineNo">055</span><a name="line.55"></a> +<span class="sourceLineNo">056</span> private final PriorityBlockingQueue<Path> logQueue;<a name="line.56"></a> +<span class="sourceLineNo">057</span> private final FileSystem fs;<a name="line.57"></a> +<span class="sourceLineNo">058</span> private final Configuration conf;<a name="line.58"></a> +<span class="sourceLineNo">059</span> private final WALEntryFilter filter;<a name="line.59"></a> +<span class="sourceLineNo">060</span> private final ReplicationSource source;<a name="line.60"></a> +<span class="sourceLineNo">061</span><a name="line.61"></a> +<span class="sourceLineNo">062</span> protected final BlockingQueue<WALEntryBatch> entryBatchQueue;<a name="line.62"></a> +<span class="sourceLineNo">063</span> // max (heap) size of each batch - multiply by number of batches in queue to get total<a name="line.63"></a> +<span class="sourceLineNo">064</span> private final long replicationBatchSizeCapacity;<a name="line.64"></a> +<span class="sourceLineNo">065</span> // max count of each batch - multiply by number of batches in queue to get total<a name="line.65"></a> +<span class="sourceLineNo">066</span> private final int replicationBatchCountCapacity;<a name="line.66"></a> +<span class="sourceLineNo">067</span> // position in the WAL to start reading at<a name="line.67"></a> +<span class="sourceLineNo">068</span> private long currentPosition;<a name="line.68"></a> +<span class="sourceLineNo">069</span> private final long sleepForRetries;<a name="line.69"></a> +<span class="sourceLineNo">070</span> private final int maxRetriesMultiplier;<a name="line.70"></a> +<span class="sourceLineNo">071</span> private final boolean eofAutoRecovery;<a name="line.71"></a> +<span class="sourceLineNo">072</span><a name="line.72"></a> +<span class="sourceLineNo">073</span> //Indicates whether this particular worker is running<a name="line.73"></a> +<span class="sourceLineNo">074</span> private boolean isReaderRunning = true;<a name="line.74"></a> +<span class="sourceLineNo">075</span><a name="line.75"></a> +<span class="sourceLineNo">076</span> private AtomicLong totalBufferUsed;<a name="line.76"></a> +<span class="sourceLineNo">077</span> private long totalBufferQuota;<a name="line.77"></a> +<span class="sourceLineNo">078</span><a name="line.78"></a> +<span class="sourceLineNo">079</span> /**<a name="line.79"></a> +<span class="sourceLineNo">080</span> * Creates a reader worker for a given WAL queue. Reads WAL entries off a given queue, batches the<a name="line.80"></a> +<span class="sourceLineNo">081</span> * entries, and puts them on a batch queue.<a name="line.81"></a> +<span class="sourceLineNo">082</span> * @param fs the files system to use<a name="line.82"></a> +<span class="sourceLineNo">083</span> * @param conf configuration to use<a name="line.83"></a> +<span class="sourceLineNo">084</span> * @param logQueue The WAL queue to read off of<a name="line.84"></a> +<span class="sourceLineNo">085</span> * @param startPosition position in the first WAL to start reading from<a name="line.85"></a> +<span class="sourceLineNo">086</span> * @param filter The filter to use while reading<a name="line.86"></a> +<span class="sourceLineNo">087</span> * @param source replication source<a name="line.87"></a> +<span class="sourceLineNo">088</span> */<a name="line.88"></a> +<span class="sourceLineNo">089</span> public ReplicationSourceWALReader(FileSystem fs, Configuration conf,<a name="line.89"></a> +<span class="sourceLineNo">090</span> PriorityBlockingQueue<Path> logQueue, long startPosition, WALEntryFilter filter,<a name="line.90"></a> +<span class="sourceLineNo">091</span> ReplicationSource source) {<a name="line.91"></a> +<span class="sourceLineNo">092</span> this.logQueue = logQueue;<a name="line.92"></a> +<span class="sourceLineNo">093</span> this.currentPosition = startPosition;<a name="line.93"></a> +<span class="sourceLineNo">094</span> this.fs = fs;<a name="line.94"></a> +<span class="sourceLineNo">095</span> this.conf = conf;<a name="line.95"></a> +<span class="sourceLineNo">096</span> this.filter = filter;<a name="line.96"></a> +<span class="sourceLineNo">097</span> this.source = source;<a name="line.97"></a> +<span class="sourceLineNo">098</span> this.replicationBatchSizeCapacity =<a name="line.98"></a> +<span class="sourceLineNo">099</span> this.conf.getLong("replication.source.size.capacity", 1024 * 1024 * 64);<a name="line.99"></a> +<span class="sourceLineNo">100</span> this.replicationBatchCountCapacity = this.conf.getInt("replication.source.nb.capacity", 25000);<a name="line.100"></a> +<span class="sourceLineNo">101</span> // memory used will be batchSizeCapacity * (nb.batches + 1)<a name="line.101"></a> +<span class="sourceLineNo">102</span> // the +1 is for the current thread reading before placing onto the queue<a name="line.102"></a> +<span class="sourceLineNo">103</span> int batchCount = conf.getInt("replication.source.nb.batches", 1);<a name="line.103"></a> +<span class="sourceLineNo">104</span> this.totalBufferUsed = source.getSourceManager().getTotalBufferUsed();<a name="line.104"></a> +<span class="sourceLineNo">105</span> this.totalBufferQuota = conf.getLong(HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_KEY,<a name="line.105"></a> +<span class="sourceLineNo">106</span> HConstants.REPLICATION_SOURCE_TOTAL_BUFFER_DFAULT);<a name="line.106"></a> +<span class="sourceLineNo">107</span> this.sleepForRetries =<a name="line.107"></a> +<span class="sourceLineNo">108</span> this.conf.getLong("replication.source.sleepforretries", 1000); // 1 second<a name="line.108"></a> +<span class="sourceLineNo">109</span> this.maxRetriesMultiplier =<a name="line.109"></a> +<span class="sourceLineNo">110</span> this.conf.getInt("replication.source.maxretriesmultiplier", 300); // 5 minutes @ 1 sec per<a name="line.110"></a> +<span class="sourceLineNo">111</span> this.eofAutoRecovery = conf.getBoolean("replication.source.eof.autorecovery", false);<a name="line.111"></a> +<span class="sourceLineNo">112</span> this.entryBatchQueue = new LinkedBlockingQueue<>(batchCount);<a name="line.112"></a> +<span class="sourceLineNo">113</span> LOG.info("peerClusterZnode=" + source.getQueueId()<a name="line.113"></a> +<span class="sourceLineNo">114</span> + ", ReplicationSourceWALReaderThread : " + source.getPeerId()<a name="line.114"></a> +<span class="sourceLineNo">115</span> + " inited, replicationBatchSizeCapacity=" + replicationBatchSizeCapacity<a name="line.115"></a> +<span class="sourceLineNo">116</span> + ", replicationBatchCountCapacity=" + replicationBatchCountCapacity<a name="line.116"></a> +<span class="sourceLineNo">117</span> + ", replicationBatchQueueCapacity=" + batchCount);<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> @Override<a name="line.120"></a> +<span class="sourceLineNo">121</span> public void run() {<a name="line.121"></a> +<span class="sourceLineNo">122</span> int sleepMultiplier = 1;<a name="line.122"></a> +<span class="sourceLineNo">123</span> while (isReaderRunning()) { // we only loop back here if something fatal happened to our stream<a name="line.123"></a> +<span class="sourceLineNo">124</span> try (WALEntryStream entryStream =<a name="line.124"></a> +<span class="sourceLineNo">125</span> new WALEntryStream(logQueue, fs, conf, currentPosition,<a name="line.125"></a> +<span class="sourceLineNo">126</span> source.getWALFileLengthProvider(), source.getServerWALsBelongTo(),<a name="line.126"></a> +<span class="sourceLineNo">127</span> source.getSourceMetrics())) {<a name="line.127"></a> +<span class="sourceLineNo">128</span> while (isReaderRunning()) { // loop here to keep reusing stream while we can<a name="line.128"></a> +<span class="sourceLineNo">129</span> if (!checkQuota()) {<a name="line.129"></a> +<span class="sourceLineNo">130</span> continue;<a name="line.130"></a> +<span class="sourceLineNo">131</span> }<a name="line.131"></a> +<span class="sourceLineNo">132</span> WALEntryBatch batch = readWALEntries(entryStream);<a name="line.132"></a> +<span class="sourceLineNo">133</span> if (batch != null) {<a name="line.133"></a> +<span class="sourceLineNo">134</span> // need to propagate the batch even it has no entries since it may carry the last<a name="line.134"></a> +<span class="sourceLineNo">135</span> // sequence id information for serial replication.<a name="line.135"></a> +<span class="sourceLineNo">136</span> LOG.trace("Read {} WAL entries eligible for replication", batch.getNbEntries());<a name="line.136"></a> +<span class="sourceLineNo">137</span> entryBatchQueue.put(batch);<a name="line.137"></a> +<span class="sourceLineNo">138</span> sleepMultiplier = 1;<a name="line.138"></a> +<span class="sourceLineNo">139</span> } else { // got no entries and didn't advance position in WAL<a name="line.139"></a> +<span class="sourceLineNo">140</span> handleEmptyWALEntryBatch(entryStream.getCurrentPath());<a name="line.140"></a> +<span class="sourceLineNo">141</span> }<a name="line.141"></a> +<span class="sourceLineNo">142</span> currentPosition = entryStream.getPosition();<a name="line.142"></a> +<span class="sourceLineNo">143</span> entryStream.reset(); // reuse stream<a name="line.143"></a> +<span class="sourceLineNo">144</span> }<a name="line.144"></a> +<span class="sourceLineNo">145</span> } catch (IOException e) { // stream related<a name="line.145"></a> +<span class="sourceLineNo">146</span> if (sleepMultiplier < maxRetriesMultiplier) {<a name="line.146"></a> +<span class="sourceLineNo">147</span> LOG.debug("Failed to read stream of replication entries: " + e);<a name="line.147"></a> +<span class="sourceLineNo">148</span> sleepMultiplier++;<a name="line.148"></a> +<span class="sourceLineNo">149</span> } else {<a name="line.149"></a> +<span class="sourceLineNo">150</span> LOG.error("Failed to read stream of replication entries", e);<a name="line.150"></a> +<span class="sourceLineNo">151</span> handleEofException(e);<a name="line.151"></a> +<span class="sourceLineNo">152</span> }<a name="line.152"></a> +<span class="sourceLineNo">153</span> Threads.sleep(sleepForRetries * sleepMultiplier);<a name="line.153"></a> +<span class="sourceLineNo">154</span> } catch (InterruptedException e) {<a name="line.154"></a> +<span class="sourceLineNo">155</span> LOG.trace("Interrupted while sleeping between WAL reads");<a name="line.155"></a> +<span class="sourceLineNo">156</span> Thread.currentThread().interrupt();<a name="line.156"></a> +<span class="sourceLineNo">157</span> }<a name="line.157"></a> +<span class="sourceLineNo">158</span> }<a name="line.158"></a> +<span class="sourceLineNo">159</span> }<a name="line.159"></a> +<span class="sourceLineNo">160</span><a name="line.160"></a> +<span class="sourceLineNo">161</span> // returns true if we reach the size limit for batch, i.e, we need to finish the batch and return.<a name="line.161"></a> +<span class="sourceLineNo">162</span> protected final boolean addEntryToBatch(WALEntryBatch batch, Entry entry) {<a name="line.162"></a> +<span class="sourceLineNo">163</span> WALEdit edit = entry.getEdit();<a name="line.163"></a> +<span class="sourceLineNo">164</span> if (edit == null || edit.isEmpty()) {<a name="line.164"></a> +<span class="sourceLineNo">165</span> return false;<a name="line.165"></a> +<span class="sourceLineNo">166</span> }<a name="line.166"></a> +<span class="sourceLineNo">167</span> long entrySize = getEntrySize(entry);<a name="line.167"></a> +<span class="sourceLineNo">168</span> batch.addEntry(entry);<a name="line.168"></a> +<span class="sourceLineNo">169</span> updateBatchStats(batch, entry, entrySize);<a name="line.169"></a> +<span class="sourceLineNo">170</span> boolean totalBufferTooLarge = acquireBufferQuota(entrySize);<a name="line.170"></a> +<span class="sourceLineNo">171</span> // Stop if too many entries or too big<a name="line.171"></a> +<span class="sourceLineNo">172</span> return totalBufferTooLarge || batch.getHeapSize() >= replicationBatchSizeCapacity ||<a name="line.172"></a> +<span class="sourceLineNo">173</span> batch.getNbEntries() >= replicationBatchCountCapacity;<a name="line.173"></a> +<span class="sourceLineNo">174</span> }<a name="line.174"></a> +<span class="sourceLineNo">175</span><a name="line.175"></a> +<span class="sourceLineNo">176</span> protected WALEntryBatch readWALEntries(WALEntryStream entryStream)<a name="line.176"></a> +<span class="sourceLineNo">177</span> throws IOException, InterruptedException {<a name="line.177"></a> +<span class="sourceLineNo">178</span> if (!entryStream.hasNext()) {<a name="line.178"></a> +<span class="sourceLineNo">179</span> return null;<a name="line.179"></a> +<span class="sourceLineNo">180</span> }<a name="line.180"></a> +<span class="sourceLineNo">181</span> WALEntryBatch batch = createBatch(entryStream);<a name="line.181"></a> +<span class="sourceLineNo">182</span> do {<a name="line.182"></a> +<span class="sourceLineNo">183</span> Entry entry = entryStream.next();<a name="line.183"></a> +<span class="sourceLineNo">184</span> batch.setLastWalPosition(entryStream.getPosition());<a name="line.184"></a> +<span class="sourceLineNo">185</span> entry = filterEntry(entry);<a name="line.185"></a> +<span class="sourceLineNo">186</span> if (entry != null) {<a name="line.186"></a> +<span class="sourceLineNo">187</span> if (addEntryToBatch(batch, entry)) {<a name="line.187"></a> +<span class="sourceLineNo">188</span> break;<a name="line.188"></a> +<span class="sourceLineNo">189</span> }<a name="line.189"></a> +<span class="sourceLineNo">190</span> }<a name="line.190"></a> +<span class="sourceLineNo">191</span> } while (entryStream.hasNext());<a name="line.191"></a> +<span class="sourceLineNo">192</span> return batch;<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> protected void handleEmptyWALEntryBatch(Path currentPath) throws InterruptedException {<a name="line.195"></a> +<span class="sourceLineNo">196</span> LOG.trace("Didn't read any new entries from WAL");<a name="line.196"></a> +<span class="sourceLineNo">197</span> Thread.sleep(sleepForRetries);<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> // if we get an EOF due to a zero-length log, and there are other logs in queue<a name="line.200"></a> +<span class="sourceLineNo">201</span> // (highly likely we've closed the current log), we've hit the max retries, and autorecovery is<a name="line.201"></a> +<span class="sourceLineNo">202</span> // enabled, then dump the log<a name="line.202"></a> +<span class="sourceLineNo">203</span> private void handleEofException(IOException e) {<a name="line.203"></a> +<span class="sourceLineNo">204</span> if ((e instanceof EOFException || e.getCause() instanceof EOFException) &&<a name="line.204"></a> +<span class="sourceLineNo">205</span> logQueue.size() > 1 && this.eofAutoRecovery) {<a name="line.205"></a> +<span class="sourceLineNo">206</span> try {<a name="line.206"></a> +<span class="sourceLineNo">207</span> if (fs.getFileStatus(logQueue.peek()).getLen() == 0) {<a name="line.207"></a> +<span class="sourceLineNo">208</span> LOG.warn("Forcing removal of 0 length log in queue: " + logQueue.peek());<a name="line.208"></a> +<span class="sourceLineNo">209</span> logQueue.remove();<a name="line.209"></a> +<span class="sourceLineNo">210</span> currentPosition = 0;<a name="line.210"></a> +<span class="sourceLineNo">211</span> }<a name="line.211"></a> +<span class="sourceLineNo">212</span> } catch (IOException ioe) {<a name="line.212"></a> +<span class="sourceLineNo">213</span> LOG.warn("Couldn't get file length information about log " + logQueue.peek());<a name="line.213"></a> +<span class="sourceLineNo">214</span> }<a name="line.214"></a> +<span class="sourceLineNo">215</span> }<a name="line.215"></a> +<span class="sourceLineNo">216</span> }<a name="line.216"></a> +<span class="sourceLineNo">217</span><a name="line.217"></a> +<span class="sourceLineNo">218</span> public Path getCurrentPath() {<a name="line.218"></a> +<span class="sourceLineNo">219</span> // if we've read some WAL entries, get the Path we read from<a name="line.219"></a> +<span class="sourceLineNo">220</span> WALEntryBatch batchQueueHead = entryBatchQueue.peek();<a name="line.220"></a> +<span class="sourceLineNo">221</span> if (batchQueueHead != null) {<a name="line.221"></a> +<span class="sourceLineNo">222</span> return batchQueueHead.getLastWalPath();<a name="line.222"></a> +<span class="sourceLineNo">223</span> }<a name="line.223"></a> +<span class="sourceLineNo">224</span> // otherwise, we must be currently reading from the head of the log queue<a name="line.224"></a> +<span class="sourceLineNo">225</span> return logQueue.peek();<a name="line.225"></a> +<span class="sourceLineNo">226</span> }<a name="line.226"></a> +<span class="sourceLineNo">227</span><a name="line.227"></a> +<span class="sourceLineNo">228</span> //returns false if we've already exceeded the global quota<a name="line.228"></a> +<span class="sourceLineNo">229</span> private boolean checkQuota() {<a name="line.229"></a> +<span class="sourceLineNo">230</span> // try not to go over total quota<a name="line.230"></a> +<span class="sourceLineNo">231</span> if (totalBufferUsed.get() > totalBufferQuota) {<a name="line.231"></a> +<span class="sourceLineNo">232</span> Threads.sleep(sleepForRetries);<a name="line.232"></a> +<span class="sourceLineNo">233</span> return false;<a name="line.233"></a> +<span class="sourceLineNo">234</span> }<a name="line.234"></a> +<span class="sourceLineNo">235</span> return true;<a name="line.235"></a> +<span class="sourceLineNo">236</span> }<a name="line.236"></a> +<span class="sourceLineNo">237</span><a name="line.237"></a> +<span class="sourceLineNo">238</span> protected final WALEntryBatch createBatch(WALEntryStream entryStream) {<a name="line.238"></a> +<span class="sourceLineNo">239</span> return new WALEntryBatch(replicationBatchCountCapacity, entryStream.getCurrentPath());<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> protected final Entry filterEntry(Entry entry) {<a name="line.242"></a> +<span class="sourceLineNo">243</span> Entry filtered = filter.filter(entry);<a name="line.243"></a> +<span class="sourceLineNo">244</span> if (entry != null && filtered == null) {<a name="line.244"></a> +<span class="sourceLineNo">245</span> source.getSourceMetrics().incrLogEditsFiltered();<a name="line.245"></a> +<span class="sourceLineNo">246</span> }<a name="line.246"></a> +<span class="sourceLineNo">247</span> return filtered;<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> /**<a name="line.250"></a> +<span class="sourceLineNo">251</span> * Retrieves the next batch of WAL entries from the queue, waiting up to the specified time for a<a name="line.251"></a> +<span class="sourceLineNo">252</span> * batch to become available<a name="line.252"></a> +<span class="sourceLineNo">253</span> * @return A batch of entries, along with the position in the log after reading the batch<a name="line.253"></a> +<span class="sourceLineNo">254</span> * @throws InterruptedException if interrupted while waiting<a name="line.254"></a> +<span class="sourceLineNo">255</span> */<a name="line.255"></a> +<span class="sourceLineNo">256</span> public WALEntryBatch take() throws InterruptedException {<a name="line.256"></a> +<span class="sourceLineNo">257</span> return entryBatchQueue.take();<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> private long getEntrySize(Entry entry) {<a name="line.260"></a> +<span class="sourceLineNo">261</span> WALEdit edit = entry.getEdit();<a name="line.261"></a> +<span class="sourceLineNo">262</span> return edit.heapSize() + calculateTotalSizeOfStoreFiles(edit);<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> private void updateBatchStats(WALEntryBatch batch, Entry entry, long entrySize) {<a name="line.265"></a> +<span class="sourceLineNo">266</span> WALEdit edit = entry.getEdit();<a name="line.266"></a> +<span class="sourceLineNo">267</span> batch.incrementHeapSize(entrySize);<a name="line.267"></a> +<span class="sourceLineNo">268</span> Pair<Integer, Integer> nbRowsAndHFiles = countDistinctRowKeysAndHFiles(edit);<a name="line.268"></a> +<span class="sourceLineNo">269</span> batch.incrementNbRowKeys(nbRowsAndHFiles.getFirst());<a name="line.269"></a> +<span class="sourceLineNo">270</span> batch.incrementNbHFiles(nbRowsAndHFiles.getSecond());<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> /**<a name="line.273"></a> +<span class="sourceLineNo">274</span> * Count the number of different row keys in the given edit because of mini-batching. We assume<a name="line.274"></a> +<span class="sourceLineNo">275</span> * that there's at least one Cell in the WALEdit.<a name="line.275"></a> +<span class="sourceLineNo">276</span> * @param edit edit to count row keys from<a name="line.276"></a> +<span class="sourceLineNo">277</span> * @return number of different row keys and HFiles<a name="line.277"></a> +<span class="sourceLineNo">278</span> */<a name="line.278"></a> +<span class="sourceLineNo">279</span> private Pair<Integer, Integer> countDistinctRowKeysAndHFiles(WALEdit edit) {<a name="line.279"></a> +<span class="sourceLineNo">280</span> List<Cell> cells = edit.getCells();<a name="line.280"></a> +<span class="sourceLineNo">281</span> int distinctRowKeys = 1;<a name="line.281"></a> +<span class="sourceLineNo">282</span> int totalHFileEntries = 0;<a name="line.282"></a> +<span class="sourceLineNo">283</span> Cell lastCell = cells.get(0);<a name="line.283"></a> +<span class="sourceLineNo">284</span><a name="line.284"></a> +<span class="sourceLineNo">285</span> int totalCells = edit.size();<a name="line.285"></a> +<span class="sourceLineNo">286</span> for (int i = 0; i < totalCells; i++) {<a name="line.286"></a> +<span class="sourceLineNo">287</span> // Count HFiles to be replicated<a name="line.287"></a> +<span class="sourceLineNo">288</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.288"></a> +<span class="sourceLineNo">289</span> try {<a name="line.289"></a> +<span class="sourceLineNo">290</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.290"></a> +<span class="sourceLineNo">291</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.291"></a> +<span class="sourceLineNo">292</span> int totalStores = stores.size();<a name="line.292"></a> +<span class="sourceLineNo">293</span> for (int j = 0; j < totalStores; j++) {<a name="line.293"></a> +<span class="sourceLineNo">294</span> totalHFileEntries += stores.get(j).getStoreFileList().size();<a name="line.294"></a> +<span class="sourceLineNo">295</span> }<a name="line.295"></a> +<span class="sourceLineNo">296</span> } catch (IOException e) {<a name="line.296"></a> +<span class="sourceLineNo">297</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.297"></a> +<span class="sourceLineNo">298</span> + "Then its hfiles count will not be added into metric.");<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><a name="line.301"></a> +<span class="sourceLineNo">302</span> if (!CellUtil.matchingRows(cells.get(i), lastCell)) {<a name="line.302"></a> +<span class="sourceLineNo">303</span> distinctRowKeys++;<a name="line.303"></a> +<span class="sourceLineNo">304</span> }<a name="line.304"></a> +<span class="sourceLineNo">305</span> lastCell = cells.get(i);<a name="line.305"></a> +<span class="sourceLineNo">306</span> }<a name="line.306"></a> +<span class="sourceLineNo">307</span><a name="line.307"></a> +<span class="sourceLineNo">308</span> Pair<Integer, Integer> result = new Pair<>(distinctRowKeys, totalHFileEntries);<a name="line.308"></a> +<span class="sourceLineNo">309</span> return result;<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> private void updateBatchStats(WALEntryBatch batch, Entry entry, long entrySize) {<a name="line.312"></a> -<span class="sourceLineNo">313</span> WALEdit edit = entry.getEdit();<a name="line.313"></a> -<span class="sourceLineNo">314</span> batch.incrementHeapSize(entrySize);<a name="line.314"></a> -<span class="sourceLineNo">315</span> Pair<Integer, Integer> nbRowsAndHFiles = countDistinctRowKeysAndHFiles(edit);<a name="line.315"></a> -<span class="sourceLineNo">316</span> batch.incrementNbRowKeys(nbRowsAndHFiles.getFirst());<a name="line.316"></a> -<span class="sourceLineNo">317</span> batch.incrementNbHFiles(nbRowsAndHFiles.getSecond());<a name="line.317"></a> -<span class="sourceLineNo">318</span> }<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> * Count the number of different row keys in the given edit because of mini-batching. We assume<a name="line.321"></a> -<span class="sourceLineNo">322</span> * that there's at least one Cell in the WALEdit.<a name="line.322"></a> -<span class="sourceLineNo">323</span> * @param edit edit to count row keys from<a name="line.323"></a> -<span class="sourceLineNo">324</span> * @return number of different row keys and HFiles<a name="line.324"></a> -<span class="sourceLineNo">325</span> */<a name="line.325"></a> -<span class="sourceLineNo">326</span> private Pair<Integer, Integer> countDistinctRowKeysAndHFiles(WALEdit edit) {<a name="line.326"></a> -<span class="sourceLineNo">327</span> List<Cell> cells = edit.getCells();<a name="line.327"></a> -<span class="sourceLineNo">328</span> int distinctRowKeys = 1;<a name="line.328"></a> -<span class="sourceLineNo">329</span> int totalHFileEntries = 0;<a name="line.329"></a> -<span class="sourceLineNo">330</span> Cell lastCell = cells.get(0);<a name="line.330"></a> -<span class="sourceLineNo">331</span><a name="line.331"></a> -<span class="sourceLineNo">332</span> int totalCells = edit.size();<a name="line.332"></a> -<span class="sourceLineNo">333</span> for (int i = 0; i < totalCells; i++) {<a name="line.333"></a> -<span class="sourceLineNo">334</span> // Count HFiles to be replicated<a name="line.334"></a> -<span class="sourceLineNo">335</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.335"></a> -<span class="sourceLineNo">336</span> try {<a name="line.336"></a> -<span class="sourceLineNo">337</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.337"></a> -<span class="sourceLineNo">338</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.338"></a> -<span class="sourceLineNo">339</span> int totalStores = stores.size();<a name="line.339"></a> -<span class="sourceLineNo">340</span> for (int j = 0; j < totalStores; j++) {<a name="line.340"></a> -<span class="sourceLineNo">341</span> totalHFileEntries += stores.get(j).getStoreFileList().size();<a name="line.341"></a> -<span class="sourceLineNo">342</span> }<a name="line.342"></a> -<span class="sourceLineNo">343</span> } catch (IOException e) {<a name="line.343"></a> -<span class="sourceLineNo">344</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.344"></a> -<span class="sourceLineNo">345</span> + "Then its hfiles count will not be added into metric.");<a name="line.345"></a> -<span class="sourceLineNo">346</span> }<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> if (!CellUtil.matchingRows(cells.get(i), lastCell)) {<a name="line.349"></a> -<span class="sourceLineNo">350</span> distinctRowKeys++;<a name="line.350"></a> -<span class="sourceLineNo">351</span> }<a name="line.351"></a> -<span class="sourceLineNo">352</span> lastCell = cells.get(i);<a name="line.352"></a> -<span class="sourceLineNo">353</span> }<a name="line.353"></a> -<span class="sourceLineNo">354</span><a name="line.354"></a> -<span class="sourceLineNo">355</span> Pair<Integer, Integer> result = new Pair<>(distinctRowKeys, totalHFileEntries);<a name="line.355"></a> -<span class="sourceLineNo">356</span> return result;<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> * Calculate the total size of all the store files<a name="line.360"></a> -<span class="sourceLineNo">361</span> * @param edit edit to count row keys from<a name="line.361"></a> -<span class="sourceLineNo">362</span> * @return the total size of the store files<a name="line.362"></a> -<span class="sourceLineNo">363</span> */<a name="line.363"></a> -<span class="sourceLineNo">364</span> private int calculateTotalSizeOfStoreFiles(WALEdit edit) {<a name="line.364"></a> -<span class="sourceLineNo">365</span> List<Cell> cells = edit.getCells();<a name="line.365"></a> -<span class="sourceLineNo">366</span> int totalStoreFilesSize = 0;<a name="line.366"></a> -<span class="sourceLineNo">367</span><a name="line.367"></a> -<span class="sourceLineNo">368</span> int totalCells = edit.size();<a name="line.368"></a> -<span class="sourceLineNo">369</span> for (int i = 0; i < totalCells; i++) {<a name="line.369"></a> -<span class="sourceLineNo">370</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.370"></a> -<span class="sourceLineNo">371</span> try {<a name="line.371"></a> -<span class="sourceLineNo">372</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.372"></a> -<span class="sourceLineNo">373</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.373"></a> -<span class="sourceLineNo">374</span> int totalStores = stores.size();<a name="line.374"></a> -<span class="sourceLineNo">375</span> for (int j = 0; j < totalStores; j++) {<a name="line.375"></a> -<span class="sourceLineNo">376</span> totalStoreFilesSize =<a name="line.376"></a> -<span class="sourceLineNo">377</span> (int) (totalStoreFilesSize + stores.get(j).getStoreFileSizeBytes());<a name="line.377"></a> -<span class="sourceLineNo">378</span> }<a name="line.378"></a> -<span class="sourceLineNo">379</span> } catch (IOException e) {<a name="line.379"></a> -<span class="sourceLineNo">380</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.380"></a> -<span class="sourceLineNo">381</span> + "Size of HFiles part of cell will not be considered in replication "<a name="line.381"></a> -<span class="sourceLineNo">382</span> + "request size calculation.",<a name="line.382"></a> -<span class="sourceLineNo">383</span> e);<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> return totalStoreFilesSize;<a name="line.387"></a> -<span class="sourceLineNo">388</span> }<a name="line.388"></a> -<span class="sourceLineNo">389</span><a name="line.389"></a> -<span class="sourceLineNo">390</span> /**<a name="line.390"></a> -<span class="sourceLineNo">391</span> * @param size delta size for grown buffer<a name="line.391"></a> -<span class="sourceLineNo">392</span> * @return true if we should clear buffer and push all<a name="line.392"></a> -<span class="sourceLineNo">393</span> */<a name="line.393"></a> -<span class="sourceLineNo">394</span> private boolean acquireBufferQuota(long size) {<a name="line.394"></a> -<span class="sourceLineNo">395</span> return totalBufferUsed.addAndGet(size) >= totalBufferQuota;<a name="line.395"></a> -<span class="sourceLineNo">396</span> }<a name="line.396"></a> -<span class="sourceLineNo">397</span><a name="line.397"></a> -<span class="sourceLineNo">398</span> /**<a name="line.398"></a> -<span class="sourceLineNo">399</span> * @return whether the reader thread is running<a name="line.399"></a> -<span class="sourceLineNo">400</span> */<a name="line.400"></a> -<span class="sourceLineNo">401</span> public boolean isReaderRunning() {<a name="line.401"></a> -<span class="sourceLineNo">402</span> return isReaderRunning && !isInterrupted();<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> * @param readerRunning the readerRunning to set<a name="line.406"></a> -<span class="sourceLineNo">407</span> */<a name="line.407"></a> -<span class="sourceLineNo">408</span> public void setReaderRunning(boolean readerRunning) {<a name="line.408"></a> -<span class="sourceLineNo">409</span> this.isReaderRunning = readerRunning;<a name="line.409"></a> -<span class="sourceLineNo">410</span> }<a name="line.410"></a> -<span class="sourceLineNo">411</span>}<a name="line.411"></a> +<span class="sourceLineNo">312</span> /**<a name="line.312"></a> +<span class="sourceLineNo">313</span> * Calculate the total size of all the store files<a name="line.313"></a> +<span class="sourceLineNo">314</span> * @param edit edit to count row keys from<a name="line.314"></a> +<span class="sourceLineNo">315</span> * @return the total size of the store files<a name="line.315"></a> +<span class="sourceLineNo">316</span> */<a name="line.316"></a> +<span class="sourceLineNo">317</span> private int calculateTotalSizeOfStoreFiles(WALEdit edit) {<a name="line.317"></a> +<span class="sourceLineNo">318</span> List<Cell> cells = edit.getCells();<a name="line.318"></a> +<span class="sourceLineNo">319</span> int totalStoreFilesSize = 0;<a name="line.319"></a> +<span class="sourceLineNo">320</span><a name="line.320"></a> +<span class="sourceLineNo">321</span> int totalCells = edit.size();<a name="line.321"></a> +<span class="sourceLineNo">322</span> for (int i = 0; i < totalCells; i++) {<a name="line.322"></a> +<span class="sourceLineNo">323</span> if (CellUtil.matchingQualifier(cells.get(i), WALEdit.BULK_LOAD)) {<a name="line.323"></a> +<span class="sourceLineNo">324</span> try {<a name="line.324"></a> +<span class="sourceLineNo">325</span> BulkLoadDescriptor bld = WALEdit.getBulkLoadDescriptor(cells.get(i));<a name="line.325"></a> +<span class="sourceLineNo">326</span> List<StoreDescriptor> stores = bld.getStoresList();<a name="line.326"></a> +<span class="sourceLineNo">327</span> int totalStores = stores.size();<a name="line.327"></a> +<span class="sourceLineNo">328</span> for (int j = 0; j < totalStores; j++) {<a name="line.328"></a> +<span class="sourceLineNo">329</span> totalStoreFilesSize =<a name="line.329"></a> +<span class="sourceLineNo">330</span> (int) (totalStoreFilesSize + stores.get(j).getStoreFileSizeBytes());<a name="line.330"></a> +<span class="sourceLineNo">331</span> }<a name="line.331"></a> +<span class="sourceLineNo">332</span> } catch (IOException e) {<a name="line.332"></a> +<span class="sourceLineNo">333</span> LOG.error("Failed to deserialize bulk load entry from wal edit. "<a name="line.333"></a> +<span class="sourceLineNo">334</span> + "Size of HFiles part of cell will not be considered in replication "<a name="line.334"></a> +<span class="sourceLineNo">335</span> + "request size calculation.",<a name="line.335"></a> +<span class="sourceLineNo">336</span> e);<a name="line.336"></a> +<span class="sourceLineNo">337</span> }<a name="line.337"></a> +<span class="sourceLineNo">338</span> }<a name="line.338"></a> +<span class="sourceLineNo">339</span> }<a name="line.339"></a> +<span class="sourceLineNo">340</span> return totalStoreFilesSize;<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> /**<a name="line.343"></a> +<span class="sourceLineNo">344</span> * @param size delta size for grown buffer<a name="line.344"></a> +<span class="sourceLineNo">345</span> * @return true if we should clear buffer and push all<a name="line.345"></a> +<span class="sourceLineNo">346</span> */<a name="line.346"></a> +<span class="sourceLineNo">347</span> private boolean acquireBufferQuota(long size) {<a name="line.347"></a> +<span class="sourceLineNo">348</span> return totalBufferUsed.addAndGet(size) >= totalBufferQuota;<a name="line.348"></a> +<span class="sourceLineNo">349</span> }<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> * @return whether the reader thread is running<a name="line.352"></a> +<span class="sourceLineNo">353</span> */<a name="line.353"></a> +<span class="sourceLineNo">354</span> public boolean isReaderRunning() {<a name="line.354"></a> +<span class="sourceLineNo">355</span> return isReaderRunning && !isInterrupted();<a name="line.355"></a> +<span class="sourceLineNo">356</span> }<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> * @param readerRunning the readerRunning to set<a name="line.359"></a> +<span class="sourceLineNo">360</span> */<a name="line.360"></a> +<span class="sourceLineNo">361</span> public void setReaderRunning(boolean readerRunning) {<a name="line.361"></a> +<span class="sourceLineNo">362</span> this.isReaderRunning = readerRunning;<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>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f1b07859/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/SerialReplicationSourceWALReader.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/SerialReplicationSourceWALReader.html b/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/SerialReplicationSourceWALReader.html new file mode 100644 index 0000000..ea32535 --- /dev/null +++ b/devapidocs/src-html/org/apache/hadoop/hbase/replication/regionserver/SerialReplicationSourceWALReader.html @@ -0,0 +1,184 @@ +<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd";> +<html lang="en"> +<head> +<title>Source code</title> +<link rel="stylesheet" type="text/css" href="../../../../../../../stylesheet.css" title="Style"> +</head> +<body> +<div class="sourceContainer"> +<pre><span class="sourceLineNo">001</span>/**<a name="line.1"></a> +<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a> +<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a> +<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a> +<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a> +<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a> +<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a> +<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a> +<span class="sourceLineNo">009</span> *<a name="line.9"></a> +<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a> +<span class="sourceLineNo">011</span> *<a name="line.11"></a> +<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a> +<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a> +<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a> +<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a> +<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a> +<span class="sourceLineNo">017</span> */<a name="line.17"></a> +<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.replication.regionserver;<a name="line.18"></a> +<span class="sourceLineNo">019</span><a name="line.19"></a> +<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a> +<span class="sourceLineNo">021</span>import java.util.concurrent.PriorityBlockingQueue;<a name="line.21"></a> +<span class="sourceLineNo">022</span>import org.apache.hadoop.conf.Configuration;<a name="line.22"></a> +<span class="sourceLineNo">023</span>import org.apache.hadoop.fs.FileSystem;<a name="line.23"></a> +<span class="sourceLineNo">024</span>import org.apache.hadoop.fs.Path;<a name="line.24"></a> +<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.Cell;<a name="line.25"></a> +<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.replication.WALEntryFilter;<a name="line.26"></a> +<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.27"></a> +<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<a name="line.28"></a> +<span class="sourceLineNo">029</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.29"></a> +<span class="sourceLineNo">030</span><a name="line.30"></a> +<span class="sourceLineNo">031</span>/**<a name="line.31"></a> +<span class="sourceLineNo">032</span> * WAL reader for a serial replication peer.<a name="line.32"></a> +<span class="sourceLineNo">033</span> */<a name="line.33"></a> +<span class="sourceLineNo">034</span>@InterfaceAudience.Private<a name="line.34"></a> +<span class="sourceLineNo">035</span>public class SerialReplicationSourceWALReader extends ReplicationSourceWALReader {<a name="line.35"></a> +<span class="sourceLineNo">036</span><a name="line.36"></a> +<span class="sourceLineNo">037</span> // used to store the first cell in an entry before filtering. This is because that if serial<a name="line.37"></a> +<span class="sourceLineNo">038</span> // replication is enabled, we may find out that an entry can not be pushed after filtering. And<a name="line.38"></a> +<span class="sourceLineNo">039</span> // when we try the next time, the cells maybe null since the entry has already been filtered,<a name="line.39"></a> +<span class="sourceLineNo">040</span> // especially for region event wal entries. And this can also used to determine whether we can<a name="line.40"></a> +<span class="sourceLineNo">041</span> // skip filtering.<a name="line.41"></a> +<span class="sourceLineNo">042</span> private Cell firstCellInEntryBeforeFiltering;<a name="line.42"></a> +<span class="sourceLineNo">043</span><a name="line.43"></a> +<span class="sourceLineNo">044</span> private final SerialReplicationChecker checker;<a name="line.44"></a> +<span class="sourceLineNo">045</span><a name="line.45"></a> +<span class="sourceLineNo">046</span> public SerialReplicationSourceWALReader(FileSystem fs, Configuration conf,<a name="line.46"></a> +<span class="sourceLineNo">047</span> PriorityBlockingQueue<Path> logQueue, long startPosition, WALEntryFilter filter,<a name="line.47"></a> +<span class="sourceLineNo">048</span> ReplicationSource source) {<a name="line.48"></a> +<span class="sourceLineNo">049</span> super(fs, conf, logQueue, startPosition, filter, source);<a name="line.49"></a> +<span class="sourceLineNo">050</span> checker = new SerialReplicationChecker(conf, source);<a name="line.50"></a> +<span class="sourceLineNo">051</span> }<a name="line.51"></a> +<span class="sourceLineNo">052</span><a name="line.52"></a> +<span class="sourceLineNo">053</span> @Override<a name="line.53"></a> +<span class="sourceLineNo">054</span> protected WALEntryBatch readWALEntries(WALEntryStream entryStream)<a name="line.54"></a> +<span class="sourceLineNo">055</span> throws IOException, InterruptedException {<a name="line.55"></a> +<span class="sourceLineNo">056</span> if (!entryStream.hasNext()) {<a name="line.56"></a> +<span class="sourceLineNo">057</span> return null;<a name="line.57"></a> +<span class="sourceLineNo">058</span> }<a name="line.58"></a> +<span class="sourceLineNo">059</span> long positionBefore = entryStream.getPosition();<a name="line.59"></a> +<span class="sourceLineNo">060</span> WALEntryBatch batch = createBatch(entryStream);<a name="line.60"></a> +<span class="sourceLineNo">061</span> do {<a name="line.61"></a> +<span class="sourceLineNo">062</span> Entry entry = entryStream.peek();<a name="line.62"></a> +<span class="sourceLineNo">063</span> boolean doFiltering = true;<a name="line.63"></a> +<span class="sourceLineNo">064</span> if (firstCellInEntryBeforeFiltering == null) {<a name="line.64"></a> +<span class="sourceLineNo">065</span> assert !entry.getEdit().isEmpty() : "should not write empty edits";<a name="line.65"></a> +<span class="sourceLineNo">066</span> // Used to locate the region record in meta table. In WAL we only have the table name and<a name="line.66"></a> +<span class="sourceLineNo">067</span> // encoded region name which can not be mapping to region name without scanning all the<a name="line.67"></a> +<span class="sourceLineNo">068</span> // records for a table, so we need a start key, just like what we have done at client side<a name="line.68"></a> +<span class="sourceLineNo">069</span> // when locating a region. For the markers, we will use the start key of the region as the<a name="line.69"></a> +<span class="sourceLineNo">070</span> // row key for the edit. And we need to do this before filtering since all the cells may<a name="line.70"></a> +<span class="sourceLineNo">071</span> // be filtered out, especially that for the markers.<a name="line.71"></a> +<span class="sourceLineNo">072</span> firstCellInEntryBeforeFiltering = entry.getEdit().getCells().get(0);<a name="line.72"></a> +<span class="sourceLineNo">073</span> } else {<a name="line.73"></a> +<span class="sourceLineNo">074</span> // if this is not null then we know that the entry has already been filtered.<a name="line.74"></a> +<span class="sourceLineNo">075</span> doFiltering = false;<a name="line.75"></a> +<span class="sourceLineNo">076</span> }<a name="line.76"></a> +<span class="sourceLineNo">077</span><a name="line.77"></a> +<span class="sourceLineNo">078</span> if (doFiltering) {<a name="line.78"></a> +<span class="sourceLineNo">079</span> entry = filterEntry(entry);<a name="line.79"></a> +<span class="sourceLineNo">080</span> }<a name="line.80"></a> +<span class="sourceLineNo">081</span> if (entry != null) {<a name="line.81"></a> +<span class="sourceLineNo">082</span> if (!checker.canPush(entry, firstCellInEntryBeforeFiltering)) {<a name="line.82"></a> +<span class="sourceLineNo">083</span> if (batch.getLastWalPosition() > positionBefore) {<a name="line.83"></a> +<span class="sourceLineNo">084</span> // we have something that can push, break<a name="line.84"></a> +<span class="sourceLineNo">085</span> break;<a name="line.85"></a> +<span class="sourceLineNo">086</span> } else {<a name="line.86"></a> +<span class="sourceLineNo">087</span> checker.waitUntilCanPush(entry, firstCellInEntryBeforeFiltering);<a name="line.87"></a> +<span class="sourceLineNo">088</span> }<a name="line.88"></a> +<span class="sourceLineNo">089</span> }<a name="line.89"></a> +<span class="sourceLineNo">090</span> // arrive here means we can push the entry, record the last sequence id<a name="line.90"></a> +<span class="sourceLineNo">091</span> batch.setLastSeqId(Bytes.toString(entry.getKey().getEncodedRegionName()),<a name="line.91"></a> +<span class="sourceLineNo">092</span> entry.getKey().getSequenceId());<a name="line.92"></a> +<span class="sourceLineNo">093</span> // actually remove the entry.<a name="line.93"></a> +<span class="sourceLineNo">094</span> removeEntryFromStream(entryStream, batch);<a name="line.94"></a> +<span class="sourceLineNo">095</span> if (addEntryToBatch(batch, entry)) {<a name="line.95"></a> +<span class="sourceLineNo">096</span> break;<a name="line.96"></a> +<span class="sourceLineNo">097</span> }<a name="line.97"></a> +<span class="sourceLineNo">098</span> } else {<a name="line.98"></a> +<span class="sourceLineNo">099</span> // actually remove the entry.<a name="line.99"></a> +<span class="sourceLineNo">100</span> removeEntryFromStream(entryStream, batch);<a name="line.100"></a> +<span class="sourceLineNo">101</span> }<a name="line.101"></a> +<span class="sourceLineNo">102</span> } while (entryStream.hasNext());<a name="line.102"></a> +<span class="sourceLineNo">103</span> return batch;<a name="line.103"></a> +<span class="sourceLineNo">104</span> }<a name="line.104"></a> +<span class="sourceLineNo">105</span><a name="line.105"></a> +<span class="sourceLineNo">106</span> private void removeEntryFromStream(WALEntryStream entryStream, WALEntryBatch batch)<a name="line.106"></a> +<span class="sourceLineNo">107</span> throws IOException {<a name="line.107"></a> +<span class="sourceLineNo">108</span> entryStream.next();<a name="line.108"></a> +<span class="sourceLineNo">109</span> firstCellInEntryBeforeFiltering = null;<a name="line.109"></a> +<span class="sourceLineNo">110</span> batch.setLastWalPosition(entryStream.getPosition());<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> + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +</pre> +</div> +</body> +</html> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f1b07859/export_control.html ---------------------------------------------------------------------- diff --git a/export_control.html b/export_control.html index cf2dce7..f0bea8f 100644 --- a/export_control.html +++ b/export_control.html @@ -7,7 +7,7 @@ <head> <meta charset="UTF-8" /> <meta name="viewport" content="width=device-width, initial-scale=1.0" /> - <meta name="Date-Revision-yyyymmdd" content="20180311" /> + <meta name="Date-Revision-yyyymmdd" content="20180312" /> <meta http-equiv="Content-Language" content="en" /> <title>Apache HBase – Export Control @@ -324,7 +324,7 @@ for more details.</p> <a href="https://www.apache.org/";>The Apache Software Foundation</a>. All rights reserved. - <li id="publishDate" class="pull-right">Last Published: 2018-03-11</li> + <li id="publishDate" class="pull-right">Last Published: 2018-03-12</li> </p> </div> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f1b07859/index.html ---------------------------------------------------------------------- diff --git a/index.html b/index.html index 8aa3913..a22852e 100644 --- a/index.html +++ b/index.html @@ -7,7 +7,7 @@ <head> <meta charset="UTF-8" /> <meta name="viewport" content="width=device-width, initial-scale=1.0" /> - <meta name="Date-Revision-yyyymmdd" content="20180311" /> + <meta name="Date-Revision-yyyymmdd" content="20180312" /> <meta http-equiv="Content-Language" content="en" /> <title>Apache HBase – Apache HBase⢠Home</title> <link rel="stylesheet" href="./css/apache-maven-fluido-1.5-HBASE.min.css" /> @@ -426,7 +426,7 @@ Apache HBase is an open-source, distributed, versioned, non-relational database <a href="https://www.apache.org/";>The Apache Software Foundation</a>. All rights reserved. - <li id="publishDate" class="pull-right">Last Published: 2018-03-11</li> + <li id="publishDate" class="pull-right">Last Published: 2018-03-12</li> </p> </div> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f1b07859/integration.html ---------------------------------------------------------------------- diff --git a/integration.html b/integration.html index dea0511..fe23b08 100644 --- a/integration.html +++ b/integration.html @@ -7,7 +7,7 @@ <head> <meta charset="UTF-8" /> <meta name="viewport" content="width=device-width, initial-scale=1.0" /> - <meta name="Date-Revision-yyyymmdd" content="20180311" /> + <meta name="Date-Revision-yyyymmdd" content="20180312" /> <meta http-equiv="Content-Language" content="en" /> <title>Apache HBase – CI Management</title> <link rel="stylesheet" href="./css/apache-maven-fluido-1.5-HBASE.min.css" /> @@ -284,7 +284,7 @@ <a href="https://www.apache.org/";>The Apache Software Foundation</a>. All rights reserved. - <li id="publishDate" class="pull-right">Last Published: 2018-03-11</li> + <li id="publishDate" class="pull-right">Last Published: 2018-03-12</li> </p> </div> http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f1b07859/issue-tracking.html ---------------------------------------------------------------------- diff --git a/issue-tracking.html b/issue-tracking.html index 568708d..2dc77ba 100644 --- a/issue-tracking.html +++ b/issue-tracking.html @@ -7,7 +7,7 @@ <head> <meta charset="UTF-8" /> <meta name="viewport" content="width=device-width, initial-scale=1.0" /> - <meta name="Date-Revision-yyyymmdd" content="20180311" /> + <meta name="Date-Revision-yyyymmdd" content="20180312" /> <meta http-equiv="Content-Language" content="en" /> <title>Apache HBase – Issue Management</title> <link rel="stylesheet" href="./css/apache-maven-fluido-1.5-HBASE.min.css" /> @@ -281,7 +281,7 @@ <a href="https://www.apache.org/";>The Apache Software Foundation</a>. All rights reserved. - <li id="publishDate" class="pull-right">Last Published: 2018-03-11</li> + <li id="publishDate" class="pull-right">Last Published: 2018-03-12</li> </p> </div>
