Repository: hbase-site Updated Branches: refs/heads/asf-site c41a1fcb4 -> e0fb1fdea
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/e0fb1fde/devapidocs/src-html/org/apache/hadoop/hbase/master/cleaner/CleanerChore.Action.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/master/cleaner/CleanerChore.Action.html b/devapidocs/src-html/org/apache/hadoop/hbase/master/cleaner/CleanerChore.Action.html index c878296..c14b16b 100644 --- a/devapidocs/src-html/org/apache/hadoop/hbase/master/cleaner/CleanerChore.Action.html +++ b/devapidocs/src-html/org/apache/hadoop/hbase/master/cleaner/CleanerChore.Action.html @@ -33,478 +33,540 @@ <span class="sourceLineNo">025</span>import java.util.Map;<a name="line.25"></a> <span class="sourceLineNo">026</span>import java.util.concurrent.ExecutionException;<a name="line.26"></a> <span class="sourceLineNo">027</span>import java.util.concurrent.ForkJoinPool;<a name="line.27"></a> -<span class="sourceLineNo">028</span>import java.util.concurrent.RecursiveTask;<a name="line.28"></a> -<span class="sourceLineNo">029</span>import java.util.concurrent.atomic.AtomicBoolean;<a name="line.29"></a> -<span class="sourceLineNo">030</span>import org.apache.hadoop.conf.Configuration;<a name="line.30"></a> -<span class="sourceLineNo">031</span>import org.apache.hadoop.fs.FileStatus;<a name="line.31"></a> -<span class="sourceLineNo">032</span>import org.apache.hadoop.fs.FileSystem;<a name="line.32"></a> -<span class="sourceLineNo">033</span>import org.apache.hadoop.fs.Path;<a name="line.33"></a> -<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.ScheduledChore;<a name="line.34"></a> -<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.Stoppable;<a name="line.35"></a> -<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.conf.ConfigurationObserver;<a name="line.36"></a> -<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.util.FSUtils;<a name="line.37"></a> -<span class="sourceLineNo">038</span>import org.apache.hadoop.ipc.RemoteException;<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.slf4j.Logger;<a name="line.40"></a> -<span class="sourceLineNo">041</span>import org.slf4j.LoggerFactory;<a name="line.41"></a> -<span class="sourceLineNo">042</span><a name="line.42"></a> -<span class="sourceLineNo">043</span>import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.43"></a> -<span class="sourceLineNo">044</span>import org.apache.hbase.thirdparty.com.google.common.base.Predicate;<a name="line.44"></a> -<span class="sourceLineNo">045</span>import org.apache.hbase.thirdparty.com.google.common.collect.ImmutableSet;<a name="line.45"></a> -<span class="sourceLineNo">046</span>import org.apache.hbase.thirdparty.com.google.common.collect.Iterables;<a name="line.46"></a> -<span class="sourceLineNo">047</span>import org.apache.hbase.thirdparty.com.google.common.collect.Lists;<a name="line.47"></a> -<span class="sourceLineNo">048</span><a name="line.48"></a> -<span class="sourceLineNo">049</span>/**<a name="line.49"></a> -<span class="sourceLineNo">050</span> * Abstract Cleaner that uses a chain of delegates to clean a directory of files<a name="line.50"></a> -<span class="sourceLineNo">051</span> * @param <T> Cleaner delegate class that is dynamically loaded from configuration<a name="line.51"></a> -<span class="sourceLineNo">052</span> */<a name="line.52"></a> -<span class="sourceLineNo">053</span>@edu.umd.cs.findbugs.annotations.SuppressWarnings(value="ST_WRITE_TO_STATIC_FROM_INSTANCE_METHOD",<a name="line.53"></a> -<span class="sourceLineNo">054</span> justification="TODO: Fix. It is wonky have static pool initialized from instance")<a name="line.54"></a> -<span class="sourceLineNo">055</span>@InterfaceAudience.Private<a name="line.55"></a> -<span class="sourceLineNo">056</span>public abstract class CleanerChore<T extends FileCleanerDelegate> extends ScheduledChore<a name="line.56"></a> -<span class="sourceLineNo">057</span> implements ConfigurationObserver {<a name="line.57"></a> -<span class="sourceLineNo">058</span><a name="line.58"></a> -<span class="sourceLineNo">059</span> private static final Logger LOG = LoggerFactory.getLogger(CleanerChore.class);<a name="line.59"></a> -<span class="sourceLineNo">060</span> private static final int AVAIL_PROCESSORS = Runtime.getRuntime().availableProcessors();<a name="line.60"></a> -<span class="sourceLineNo">061</span><a name="line.61"></a> -<span class="sourceLineNo">062</span> /**<a name="line.62"></a> -<span class="sourceLineNo">063</span> * If it is an integer and >= 1, it would be the size;<a name="line.63"></a> -<span class="sourceLineNo">064</span> * if 0.0 < size <= 1.0, size would be available processors * size.<a name="line.64"></a> -<span class="sourceLineNo">065</span> * Pay attention that 1.0 is different from 1, former indicates it will use 100% of cores,<a name="line.65"></a> -<span class="sourceLineNo">066</span> * while latter will use only 1 thread for chore to scan dir.<a name="line.66"></a> -<span class="sourceLineNo">067</span> */<a name="line.67"></a> -<span class="sourceLineNo">068</span> public static final String CHORE_POOL_SIZE = "hbase.cleaner.scan.dir.concurrent.size";<a name="line.68"></a> -<span class="sourceLineNo">069</span> private static final String DEFAULT_CHORE_POOL_SIZE = "0.25";<a name="line.69"></a> -<span class="sourceLineNo">070</span><a name="line.70"></a> -<span class="sourceLineNo">071</span> // It may be waste resources for each cleaner chore own its pool,<a name="line.71"></a> -<span class="sourceLineNo">072</span> // so let's make pool for all cleaner chores.<a name="line.72"></a> -<span class="sourceLineNo">073</span> private static volatile ForkJoinPool CHOREPOOL;<a name="line.73"></a> -<span class="sourceLineNo">074</span> private static volatile int CHOREPOOLSIZE;<a name="line.74"></a> -<span class="sourceLineNo">075</span><a name="line.75"></a> -<span class="sourceLineNo">076</span> protected final FileSystem fs;<a name="line.76"></a> -<span class="sourceLineNo">077</span> private final Path oldFileDir;<a name="line.77"></a> -<span class="sourceLineNo">078</span> private final Configuration conf;<a name="line.78"></a> -<span class="sourceLineNo">079</span> protected final Map<String, Object> params;<a name="line.79"></a> -<span class="sourceLineNo">080</span> private final AtomicBoolean enabled = new AtomicBoolean(true);<a name="line.80"></a> -<span class="sourceLineNo">081</span> private final AtomicBoolean reconfig = new AtomicBoolean(false);<a name="line.81"></a> -<span class="sourceLineNo">082</span> protected List<T> cleanersChain;<a name="line.82"></a> -<span class="sourceLineNo">083</span><a name="line.83"></a> -<span class="sourceLineNo">084</span> public CleanerChore(String name, final int sleepPeriod, final Stoppable s, Configuration conf,<a name="line.84"></a> -<span class="sourceLineNo">085</span> FileSystem fs, Path oldFileDir, String confKey) {<a name="line.85"></a> -<span class="sourceLineNo">086</span> this(name, sleepPeriod, s, conf, fs, oldFileDir, confKey, null);<a name="line.86"></a> -<span class="sourceLineNo">087</span> }<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> * @param name name of the chore being run<a name="line.90"></a> -<span class="sourceLineNo">091</span> * @param sleepPeriod the period of time to sleep between each run<a name="line.91"></a> -<span class="sourceLineNo">092</span> * @param s the stopper<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 fs handle to the FS<a name="line.94"></a> -<span class="sourceLineNo">095</span> * @param oldFileDir the path to the archived files<a name="line.95"></a> -<span class="sourceLineNo">096</span> * @param confKey configuration key for the classes to instantiate<a name="line.96"></a> -<span class="sourceLineNo">097</span> * @param params members could be used in cleaner<a name="line.97"></a> -<span class="sourceLineNo">098</span> */<a name="line.98"></a> -<span class="sourceLineNo">099</span> public CleanerChore(String name, final int sleepPeriod, final Stoppable s, Configuration conf,<a name="line.99"></a> -<span class="sourceLineNo">100</span> FileSystem fs, Path oldFileDir, String confKey, Map<String, Object> params) {<a name="line.100"></a> -<span class="sourceLineNo">101</span> super(name, s, sleepPeriod);<a name="line.101"></a> -<span class="sourceLineNo">102</span> this.fs = fs;<a name="line.102"></a> -<span class="sourceLineNo">103</span> this.oldFileDir = oldFileDir;<a name="line.103"></a> -<span class="sourceLineNo">104</span> this.conf = conf;<a name="line.104"></a> -<span class="sourceLineNo">105</span> this.params = params;<a name="line.105"></a> -<span class="sourceLineNo">106</span> initCleanerChain(confKey);<a name="line.106"></a> -<span class="sourceLineNo">107</span><a name="line.107"></a> -<span class="sourceLineNo">108</span> if (CHOREPOOL == null) {<a name="line.108"></a> -<span class="sourceLineNo">109</span> String poolSize = conf.get(CHORE_POOL_SIZE, DEFAULT_CHORE_POOL_SIZE);<a name="line.109"></a> -<span class="sourceLineNo">110</span> CHOREPOOLSIZE = calculatePoolSize(poolSize);<a name="line.110"></a> -<span class="sourceLineNo">111</span> // poolSize may be 0 or 0.0 from a careless configuration,<a name="line.111"></a> -<span class="sourceLineNo">112</span> // double check to make sure.<a name="line.112"></a> -<span class="sourceLineNo">113</span> CHOREPOOLSIZE = CHOREPOOLSIZE == 0? calculatePoolSize(DEFAULT_CHORE_POOL_SIZE): CHOREPOOLSIZE;<a name="line.113"></a> -<span class="sourceLineNo">114</span> this.CHOREPOOL = new ForkJoinPool(CHOREPOOLSIZE);<a name="line.114"></a> -<span class="sourceLineNo">115</span> LOG.info("Cleaner pool size is {}", CHOREPOOLSIZE);<a name="line.115"></a> -<span class="sourceLineNo">116</span> }<a name="line.116"></a> -<span class="sourceLineNo">117</span> }<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> * Calculate size for cleaner pool.<a name="line.120"></a> -<span class="sourceLineNo">121</span> * @param poolSize size from configuration<a name="line.121"></a> -<span class="sourceLineNo">122</span> * @return size of pool after calculation<a name="line.122"></a> -<span class="sourceLineNo">123</span> */<a name="line.123"></a> -<span class="sourceLineNo">124</span> static int calculatePoolSize(String poolSize) {<a name="line.124"></a> -<span class="sourceLineNo">125</span> if (poolSize.matches("[1-9][0-9]*")) {<a name="line.125"></a> -<span class="sourceLineNo">126</span> // If poolSize is an integer, return it directly,<a name="line.126"></a> -<span class="sourceLineNo">127</span> // but upmost to the number of available processors.<a name="line.127"></a> -<span class="sourceLineNo">128</span> int size = Math.min(Integer.parseInt(poolSize), AVAIL_PROCESSORS);<a name="line.128"></a> -<span class="sourceLineNo">129</span> if (size == AVAIL_PROCESSORS) {<a name="line.129"></a> -<span class="sourceLineNo">130</span> LOG.warn("Use full core processors to scan dir, size={}", size);<a name="line.130"></a> -<span class="sourceLineNo">131</span> }<a name="line.131"></a> -<span class="sourceLineNo">132</span> return size;<a name="line.132"></a> -<span class="sourceLineNo">133</span> } else if (poolSize.matches("0.[0-9]+|1.0")) {<a name="line.133"></a> -<span class="sourceLineNo">134</span> // if poolSize is a double, return poolSize * availableProcessors;<a name="line.134"></a> -<span class="sourceLineNo">135</span> // Ensure that we always return at least one.<a name="line.135"></a> -<span class="sourceLineNo">136</span> int computedThreads = (int) (AVAIL_PROCESSORS * Double.valueOf(poolSize));<a name="line.136"></a> -<span class="sourceLineNo">137</span> if (computedThreads < 1) {<a name="line.137"></a> -<span class="sourceLineNo">138</span> LOG.debug("Computed {} threads for CleanerChore, using 1 instead", computedThreads);<a name="line.138"></a> -<span class="sourceLineNo">139</span> return 1;<a name="line.139"></a> -<span class="sourceLineNo">140</span> }<a name="line.140"></a> -<span class="sourceLineNo">141</span> return computedThreads;<a name="line.141"></a> -<span class="sourceLineNo">142</span> } else {<a name="line.142"></a> -<span class="sourceLineNo">143</span> LOG.error("Unrecognized value: " + poolSize + " for " + CHORE_POOL_SIZE +<a name="line.143"></a> -<span class="sourceLineNo">144</span> ", use default config: " + DEFAULT_CHORE_POOL_SIZE + " instead.");<a name="line.144"></a> -<span class="sourceLineNo">145</span> return calculatePoolSize(DEFAULT_CHORE_POOL_SIZE);<a name="line.145"></a> -<span class="sourceLineNo">146</span> }<a name="line.146"></a> -<span class="sourceLineNo">147</span> }<a name="line.147"></a> -<span class="sourceLineNo">148</span><a name="line.148"></a> -<span class="sourceLineNo">149</span> /**<a name="line.149"></a> -<span class="sourceLineNo">150</span> * Validate the file to see if it even belongs in the directory. If it is valid, then the file<a name="line.150"></a> -<span class="sourceLineNo">151</span> * will go through the cleaner delegates, but otherwise the file is just deleted.<a name="line.151"></a> -<span class="sourceLineNo">152</span> * @param file full {@link Path} of the file to be checked<a name="line.152"></a> -<span class="sourceLineNo">153</span> * @return <tt>true</tt> if the file is valid, <tt>false</tt> otherwise<a name="line.153"></a> -<span class="sourceLineNo">154</span> */<a name="line.154"></a> -<span class="sourceLineNo">155</span> protected abstract boolean validate(Path file);<a name="line.155"></a> -<span class="sourceLineNo">156</span><a name="line.156"></a> -<span class="sourceLineNo">157</span> /**<a name="line.157"></a> -<span class="sourceLineNo">158</span> * Instantiate and initialize all the file cleaners set in the configuration<a name="line.158"></a> -<span class="sourceLineNo">159</span> * @param confKey key to get the file cleaner classes from the configuration<a name="line.159"></a> -<span class="sourceLineNo">160</span> */<a name="line.160"></a> -<span class="sourceLineNo">161</span> private void initCleanerChain(String confKey) {<a name="line.161"></a> -<span class="sourceLineNo">162</span> this.cleanersChain = new LinkedList<>();<a name="line.162"></a> -<span class="sourceLineNo">163</span> String[] logCleaners = conf.getStrings(confKey);<a name="line.163"></a> -<span class="sourceLineNo">164</span> if (logCleaners != null) {<a name="line.164"></a> -<span class="sourceLineNo">165</span> for (String className : logCleaners) {<a name="line.165"></a> -<span class="sourceLineNo">166</span> T logCleaner = newFileCleaner(className, conf);<a name="line.166"></a> -<span class="sourceLineNo">167</span> if (logCleaner != null) {<a name="line.167"></a> -<span class="sourceLineNo">168</span> LOG.debug("Initialize cleaner={}", className);<a name="line.168"></a> -<span class="sourceLineNo">169</span> this.cleanersChain.add(logCleaner);<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> }<a name="line.172"></a> -<span class="sourceLineNo">173</span> }<a name="line.173"></a> -<span class="sourceLineNo">174</span><a name="line.174"></a> -<span class="sourceLineNo">175</span> @Override<a name="line.175"></a> -<span class="sourceLineNo">176</span> public void onConfigurationChange(Configuration conf) {<a name="line.176"></a> -<span class="sourceLineNo">177</span> int updatedSize = calculatePoolSize(conf.get(CHORE_POOL_SIZE, DEFAULT_CHORE_POOL_SIZE));<a name="line.177"></a> -<span class="sourceLineNo">178</span> if (updatedSize == CHOREPOOLSIZE) {<a name="line.178"></a> -<span class="sourceLineNo">179</span> LOG.trace("Size from configuration is same as previous={}, no need to update.", updatedSize);<a name="line.179"></a> -<span class="sourceLineNo">180</span> return;<a name="line.180"></a> -<span class="sourceLineNo">181</span> }<a name="line.181"></a> -<span class="sourceLineNo">182</span> CHOREPOOLSIZE = updatedSize;<a name="line.182"></a> -<span class="sourceLineNo">183</span> if (CHOREPOOL.getPoolSize() == 0) {<a name="line.183"></a> -<span class="sourceLineNo">184</span> // Chore does not work now, update it directly.<a name="line.184"></a> -<span class="sourceLineNo">185</span> updateChorePoolSize(updatedSize);<a name="line.185"></a> -<span class="sourceLineNo">186</span> return;<a name="line.186"></a> -<span class="sourceLineNo">187</span> }<a name="line.187"></a> -<span class="sourceLineNo">188</span> // Chore is working, update it after chore finished.<a name="line.188"></a> -<span class="sourceLineNo">189</span> reconfig.set(true);<a name="line.189"></a> -<span class="sourceLineNo">190</span> }<a name="line.190"></a> -<span class="sourceLineNo">191</span><a name="line.191"></a> -<span class="sourceLineNo">192</span> private void updateChorePoolSize(int updatedSize) {<a name="line.192"></a> -<span class="sourceLineNo">193</span> CHOREPOOL.shutdownNow();<a name="line.193"></a> -<span class="sourceLineNo">194</span> LOG.info("Update chore's pool size from {} to {}", CHOREPOOL.getParallelism(), updatedSize);<a name="line.194"></a> -<span class="sourceLineNo">195</span> CHOREPOOL = new ForkJoinPool(updatedSize);<a name="line.195"></a> -<span class="sourceLineNo">196</span> }<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 utility method to create new instances of LogCleanerDelegate based on the class name of the<a name="line.199"></a> -<span class="sourceLineNo">200</span> * LogCleanerDelegate.<a name="line.200"></a> -<span class="sourceLineNo">201</span> * @param className fully qualified class name of the LogCleanerDelegate<a name="line.201"></a> -<span class="sourceLineNo">202</span> * @param conf used configuration<a name="line.202"></a> -<span class="sourceLineNo">203</span> * @return the new instance<a name="line.203"></a> -<span class="sourceLineNo">204</span> */<a name="line.204"></a> -<span class="sourceLineNo">205</span> private T newFileCleaner(String className, Configuration conf) {<a name="line.205"></a> -<span class="sourceLineNo">206</span> try {<a name="line.206"></a> -<span class="sourceLineNo">207</span> Class<? extends FileCleanerDelegate> c = Class.forName(className).asSubclass(<a name="line.207"></a> -<span class="sourceLineNo">208</span> FileCleanerDelegate.class);<a name="line.208"></a> -<span class="sourceLineNo">209</span> @SuppressWarnings("unchecked")<a name="line.209"></a> -<span class="sourceLineNo">210</span> T cleaner = (T) c.getDeclaredConstructor().newInstance();<a name="line.210"></a> -<span class="sourceLineNo">211</span> cleaner.setConf(conf);<a name="line.211"></a> -<span class="sourceLineNo">212</span> cleaner.init(this.params);<a name="line.212"></a> -<span class="sourceLineNo">213</span> return cleaner;<a name="line.213"></a> -<span class="sourceLineNo">214</span> } catch (Exception e) {<a name="line.214"></a> -<span class="sourceLineNo">215</span> LOG.warn("Can NOT create CleanerDelegate={}", className, e);<a name="line.215"></a> -<span class="sourceLineNo">216</span> // skipping if can't instantiate<a name="line.216"></a> -<span class="sourceLineNo">217</span> return null;<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> @Override<a name="line.221"></a> -<span class="sourceLineNo">222</span> protected void chore() {<a name="line.222"></a> -<span class="sourceLineNo">223</span> if (getEnabled()) {<a name="line.223"></a> -<span class="sourceLineNo">224</span> if (runCleaner()) {<a name="line.224"></a> -<span class="sourceLineNo">225</span> LOG.debug("Cleaned all WALs under {}", oldFileDir);<a name="line.225"></a> -<span class="sourceLineNo">226</span> } else {<a name="line.226"></a> -<span class="sourceLineNo">227</span> LOG.warn("WALs outstanding under {}", oldFileDir);<a name="line.227"></a> -<span class="sourceLineNo">228</span> }<a name="line.228"></a> -<span class="sourceLineNo">229</span> // After each clean chore, checks if receives reconfigure notification while cleaning<a name="line.229"></a> -<span class="sourceLineNo">230</span> if (reconfig.compareAndSet(true, false)) {<a name="line.230"></a> -<span class="sourceLineNo">231</span> updateChorePoolSize(CHOREPOOLSIZE);<a name="line.231"></a> -<span class="sourceLineNo">232</span> }<a name="line.232"></a> -<span class="sourceLineNo">233</span> } else {<a name="line.233"></a> -<span class="sourceLineNo">234</span> LOG.debug("Cleaner chore disabled! Not cleaning.");<a name="line.234"></a> -<span class="sourceLineNo">235</span> }<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> private void preRunCleaner() {<a name="line.238"></a> -<span class="sourceLineNo">239</span> cleanersChain.forEach(FileCleanerDelegate::preClean);<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> public Boolean runCleaner() {<a name="line.242"></a> -<span class="sourceLineNo">243</span> preRunCleaner();<a name="line.243"></a> -<span class="sourceLineNo">244</span> CleanerTask task = new CleanerTask(this.oldFileDir, true);<a name="line.244"></a> -<span class="sourceLineNo">245</span> CHOREPOOL.submit(task);<a name="line.245"></a> -<span class="sourceLineNo">246</span> return task.join();<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> * Sort the given list in (descending) order of the space each element takes<a name="line.250"></a> -<span class="sourceLineNo">251</span> * @param dirs the list to sort, element in it should be directory (not file)<a name="line.251"></a> -<span class="sourceLineNo">252</span> */<a name="line.252"></a> -<span class="sourceLineNo">253</span> private void sortByConsumedSpace(List<FileStatus> dirs) {<a name="line.253"></a> -<span class="sourceLineNo">254</span> if (dirs == null || dirs.size() < 2) {<a name="line.254"></a> -<span class="sourceLineNo">255</span> // no need to sort for empty or single directory<a name="line.255"></a> -<span class="sourceLineNo">256</span> return;<a name="line.256"></a> -<span class="sourceLineNo">257</span> }<a name="line.257"></a> -<span class="sourceLineNo">258</span> dirs.sort(new Comparator<FileStatus>() {<a name="line.258"></a> -<span class="sourceLineNo">259</span> HashMap<FileStatus, Long> directorySpaces = new HashMap<>();<a name="line.259"></a> -<span class="sourceLineNo">260</span><a name="line.260"></a> -<span class="sourceLineNo">261</span> @Override<a name="line.261"></a> -<span class="sourceLineNo">262</span> public int compare(FileStatus f1, FileStatus f2) {<a name="line.262"></a> -<span class="sourceLineNo">263</span> long f1ConsumedSpace = getSpace(f1);<a name="line.263"></a> -<span class="sourceLineNo">264</span> long f2ConsumedSpace = getSpace(f2);<a name="line.264"></a> -<span class="sourceLineNo">265</span> return Long.compare(f2ConsumedSpace, f1ConsumedSpace);<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> private long getSpace(FileStatus f) {<a name="line.268"></a> -<span class="sourceLineNo">269</span> Long cached = directorySpaces.get(f);<a name="line.269"></a> -<span class="sourceLineNo">270</span> if (cached != null) {<a name="line.270"></a> -<span class="sourceLineNo">271</span> return cached;<a name="line.271"></a> -<span class="sourceLineNo">272</span> }<a name="line.272"></a> -<span class="sourceLineNo">273</span> try {<a name="line.273"></a> -<span class="sourceLineNo">274</span> long space =<a name="line.274"></a> -<span class="sourceLineNo">275</span> f.isDirectory() ? fs.getContentSummary(f.getPath()).getSpaceConsumed() : f.getLen();<a name="line.275"></a> -<span class="sourceLineNo">276</span> directorySpaces.put(f, space);<a name="line.276"></a> -<span class="sourceLineNo">277</span> return space;<a name="line.277"></a> -<span class="sourceLineNo">278</span> } catch (IOException e) {<a name="line.278"></a> -<span class="sourceLineNo">279</span> LOG.trace("Failed to get space consumed by path={}", f, e);<a name="line.279"></a> -<span class="sourceLineNo">280</span> return -1;<a name="line.280"></a> -<span class="sourceLineNo">281</span> }<a name="line.281"></a> -<span class="sourceLineNo">282</span> }<a name="line.282"></a> -<span class="sourceLineNo">283</span> });<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> /**<a name="line.286"></a> -<span class="sourceLineNo">287</span> * Run the given files through each of the cleaners to see if it should be deleted, deleting it if<a name="line.287"></a> -<span class="sourceLineNo">288</span> * necessary.<a name="line.288"></a> -<span class="sourceLineNo">289</span> * @param files List of FileStatus for the files to check (and possibly delete)<a name="line.289"></a> -<span class="sourceLineNo">290</span> * @return true iff successfully deleted all files<a name="line.290"></a> -<span class="sourceLineNo">291</span> */<a name="line.291"></a> -<span class="sourceLineNo">292</span> private boolean checkAndDeleteFiles(List<FileStatus> files) {<a name="line.292"></a> -<span class="sourceLineNo">293</span> if (files == null) {<a name="line.293"></a> -<span class="sourceLineNo">294</span> return true;<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> // first check to see if the path is valid<a name="line.297"></a> -<span class="sourceLineNo">298</span> List<FileStatus> validFiles = Lists.newArrayListWithCapacity(files.size());<a name="line.298"></a> -<span class="sourceLineNo">299</span> List<FileStatus> invalidFiles = Lists.newArrayList();<a name="line.299"></a> -<span class="sourceLineNo">300</span> for (FileStatus file : files) {<a name="line.300"></a> -<span class="sourceLineNo">301</span> if (validate(file.getPath())) {<a name="line.301"></a> -<span class="sourceLineNo">302</span> validFiles.add(file);<a name="line.302"></a> -<span class="sourceLineNo">303</span> } else {<a name="line.303"></a> -<span class="sourceLineNo">304</span> LOG.warn("Found a wrongly formatted file: " + file.getPath() + " - will delete it.");<a name="line.304"></a> -<span class="sourceLineNo">305</span> invalidFiles.add(file);<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><a name="line.308"></a> -<span class="sourceLineNo">309</span> Iterable<FileStatus> deletableValidFiles = validFiles;<a name="line.309"></a> -<span class="sourceLineNo">310</span> // check each of the cleaners for the valid files<a name="line.310"></a> -<span class="sourceLineNo">311</span> for (T cleaner : cleanersChain) {<a name="line.311"></a> -<span class="sourceLineNo">312</span> if (cleaner.isStopped() || this.getStopper().isStopped()) {<a name="line.312"></a> -<span class="sourceLineNo">313</span> LOG.warn("A file cleaner" + this.getName() + " is stopped, won't delete any more files in:"<a name="line.313"></a> -<span class="sourceLineNo">314</span> + this.oldFileDir);<a name="line.314"></a> -<span class="sourceLineNo">315</span> return false;<a name="line.315"></a> -<span class="sourceLineNo">316</span> }<a name="line.316"></a> -<span class="sourceLineNo">317</span><a name="line.317"></a> -<span class="sourceLineNo">318</span> Iterable<FileStatus> filteredFiles = cleaner.getDeletableFiles(deletableValidFiles);<a name="line.318"></a> -<span class="sourceLineNo">319</span><a name="line.319"></a> -<span class="sourceLineNo">320</span> // trace which cleaner is holding on to each file<a name="line.320"></a> -<span class="sourceLineNo">321</span> if (LOG.isTraceEnabled()) {<a name="line.321"></a> -<span class="sourceLineNo">322</span> ImmutableSet<FileStatus> filteredFileSet = ImmutableSet.copyOf(filteredFiles);<a name="line.322"></a> -<span class="sourceLineNo">323</span> for (FileStatus file : deletableValidFiles) {<a name="line.323"></a> -<span class="sourceLineNo">324</span> if (!filteredFileSet.contains(file)) {<a name="line.324"></a> -<span class="sourceLineNo">325</span> LOG.trace(file.getPath() + " is not deletable according to:" + cleaner);<a name="line.325"></a> -<span class="sourceLineNo">326</span> }<a name="line.326"></a> -<span class="sourceLineNo">327</span> }<a name="line.327"></a> +<span class="sourceLineNo">028</span>import java.util.concurrent.ForkJoinTask;<a name="line.28"></a> +<span class="sourceLineNo">029</span>import java.util.concurrent.RecursiveTask;<a name="line.29"></a> +<span class="sourceLineNo">030</span>import java.util.concurrent.atomic.AtomicBoolean;<a name="line.30"></a> +<span class="sourceLineNo">031</span>import org.apache.hadoop.conf.Configuration;<a name="line.31"></a> +<span class="sourceLineNo">032</span>import org.apache.hadoop.fs.FileStatus;<a name="line.32"></a> +<span class="sourceLineNo">033</span>import org.apache.hadoop.fs.FileSystem;<a name="line.33"></a> +<span class="sourceLineNo">034</span>import org.apache.hadoop.fs.Path;<a name="line.34"></a> +<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.ScheduledChore;<a name="line.35"></a> +<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.Stoppable;<a name="line.36"></a> +<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.conf.ConfigurationObserver;<a name="line.37"></a> +<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.util.FSUtils;<a name="line.38"></a> +<span class="sourceLineNo">039</span>import org.apache.hadoop.ipc.RemoteException;<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.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.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.44"></a> +<span class="sourceLineNo">045</span>import org.apache.hbase.thirdparty.com.google.common.base.Preconditions;<a name="line.45"></a> +<span class="sourceLineNo">046</span>import org.apache.hbase.thirdparty.com.google.common.base.Predicate;<a name="line.46"></a> +<span class="sourceLineNo">047</span>import org.apache.hbase.thirdparty.com.google.common.collect.ImmutableSet;<a name="line.47"></a> +<span class="sourceLineNo">048</span>import org.apache.hbase.thirdparty.com.google.common.collect.Iterables;<a name="line.48"></a> +<span class="sourceLineNo">049</span>import org.apache.hbase.thirdparty.com.google.common.collect.Lists;<a name="line.49"></a> +<span class="sourceLineNo">050</span><a name="line.50"></a> +<span class="sourceLineNo">051</span>/**<a name="line.51"></a> +<span class="sourceLineNo">052</span> * Abstract Cleaner that uses a chain of delegates to clean a directory of files<a name="line.52"></a> +<span class="sourceLineNo">053</span> * @param <T> Cleaner delegate class that is dynamically loaded from configuration<a name="line.53"></a> +<span class="sourceLineNo">054</span> */<a name="line.54"></a> +<span class="sourceLineNo">055</span>@edu.umd.cs.findbugs.annotations.SuppressWarnings(value="ST_WRITE_TO_STATIC_FROM_INSTANCE_METHOD",<a name="line.55"></a> +<span class="sourceLineNo">056</span> justification="Static pool will be only updated once.")<a name="line.56"></a> +<span class="sourceLineNo">057</span>@InterfaceAudience.Private<a name="line.57"></a> +<span class="sourceLineNo">058</span>public abstract class CleanerChore<T extends FileCleanerDelegate> extends ScheduledChore<a name="line.58"></a> +<span class="sourceLineNo">059</span> implements ConfigurationObserver {<a name="line.59"></a> +<span class="sourceLineNo">060</span><a name="line.60"></a> +<span class="sourceLineNo">061</span> private static final Logger LOG = LoggerFactory.getLogger(CleanerChore.class);<a name="line.61"></a> +<span class="sourceLineNo">062</span> private static final int AVAIL_PROCESSORS = Runtime.getRuntime().availableProcessors();<a name="line.62"></a> +<span class="sourceLineNo">063</span><a name="line.63"></a> +<span class="sourceLineNo">064</span> /**<a name="line.64"></a> +<span class="sourceLineNo">065</span> * If it is an integer and >= 1, it would be the size;<a name="line.65"></a> +<span class="sourceLineNo">066</span> * if 0.0 < size <= 1.0, size would be available processors * size.<a name="line.66"></a> +<span class="sourceLineNo">067</span> * Pay attention that 1.0 is different from 1, former indicates it will use 100% of cores,<a name="line.67"></a> +<span class="sourceLineNo">068</span> * while latter will use only 1 thread for chore to scan dir.<a name="line.68"></a> +<span class="sourceLineNo">069</span> */<a name="line.69"></a> +<span class="sourceLineNo">070</span> public static final String CHORE_POOL_SIZE = "hbase.cleaner.scan.dir.concurrent.size";<a name="line.70"></a> +<span class="sourceLineNo">071</span> private static final String DEFAULT_CHORE_POOL_SIZE = "0.25";<a name="line.71"></a> +<span class="sourceLineNo">072</span><a name="line.72"></a> +<span class="sourceLineNo">073</span> private static class DirScanPool {<a name="line.73"></a> +<span class="sourceLineNo">074</span> int size;<a name="line.74"></a> +<span class="sourceLineNo">075</span> ForkJoinPool pool;<a name="line.75"></a> +<span class="sourceLineNo">076</span> int cleanerLatch;<a name="line.76"></a> +<span class="sourceLineNo">077</span> AtomicBoolean reconfigNotification;<a name="line.77"></a> +<span class="sourceLineNo">078</span><a name="line.78"></a> +<span class="sourceLineNo">079</span> DirScanPool(Configuration conf) {<a name="line.79"></a> +<span class="sourceLineNo">080</span> String poolSize = conf.get(CHORE_POOL_SIZE, DEFAULT_CHORE_POOL_SIZE);<a name="line.80"></a> +<span class="sourceLineNo">081</span> size = calculatePoolSize(poolSize);<a name="line.81"></a> +<span class="sourceLineNo">082</span> // poolSize may be 0 or 0.0 from a careless configuration,<a name="line.82"></a> +<span class="sourceLineNo">083</span> // double check to make sure.<a name="line.83"></a> +<span class="sourceLineNo">084</span> size = size == 0 ? calculatePoolSize(DEFAULT_CHORE_POOL_SIZE) : size;<a name="line.84"></a> +<span class="sourceLineNo">085</span> pool = new ForkJoinPool(size);<a name="line.85"></a> +<span class="sourceLineNo">086</span> LOG.info("Cleaner pool size is {}", size);<a name="line.86"></a> +<span class="sourceLineNo">087</span> reconfigNotification = new AtomicBoolean(false);<a name="line.87"></a> +<span class="sourceLineNo">088</span> cleanerLatch = 0;<a name="line.88"></a> +<span class="sourceLineNo">089</span> }<a name="line.89"></a> +<span class="sourceLineNo">090</span><a name="line.90"></a> +<span class="sourceLineNo">091</span> /**<a name="line.91"></a> +<span class="sourceLineNo">092</span> * Checks if pool can be updated immediately.<a name="line.92"></a> +<span class="sourceLineNo">093</span> * @param conf configuration<a name="line.93"></a> +<span class="sourceLineNo">094</span> * @return true if pool can be updated immediately, false otherwise<a name="line.94"></a> +<span class="sourceLineNo">095</span> */<a name="line.95"></a> +<span class="sourceLineNo">096</span> synchronized boolean canUpdateImmediately(Configuration conf) {<a name="line.96"></a> +<span class="sourceLineNo">097</span> int newSize = calculatePoolSize(conf.get(CHORE_POOL_SIZE, DEFAULT_CHORE_POOL_SIZE));<a name="line.97"></a> +<span class="sourceLineNo">098</span> if (newSize == size) {<a name="line.98"></a> +<span class="sourceLineNo">099</span> LOG.trace("Size from configuration is same as previous={}, no need to update.", newSize);<a name="line.99"></a> +<span class="sourceLineNo">100</span> return false;<a name="line.100"></a> +<span class="sourceLineNo">101</span> }<a name="line.101"></a> +<span class="sourceLineNo">102</span> size = newSize;<a name="line.102"></a> +<span class="sourceLineNo">103</span> if (pool.getPoolSize() == 0) {<a name="line.103"></a> +<span class="sourceLineNo">104</span> // chore has no working thread.<a name="line.104"></a> +<span class="sourceLineNo">105</span> return true;<a name="line.105"></a> +<span class="sourceLineNo">106</span> }<a name="line.106"></a> +<span class="sourceLineNo">107</span> // Chore is working, update it later.<a name="line.107"></a> +<span class="sourceLineNo">108</span> reconfigNotification.set(true);<a name="line.108"></a> +<span class="sourceLineNo">109</span> return false;<a name="line.109"></a> +<span class="sourceLineNo">110</span> }<a name="line.110"></a> +<span class="sourceLineNo">111</span><a name="line.111"></a> +<span class="sourceLineNo">112</span> /**<a name="line.112"></a> +<span class="sourceLineNo">113</span> * Update pool with new size.<a name="line.113"></a> +<span class="sourceLineNo">114</span> */<a name="line.114"></a> +<span class="sourceLineNo">115</span> synchronized void updatePool(long timeout) {<a name="line.115"></a> +<span class="sourceLineNo">116</span> while (cleanerLatch != 0) {<a name="line.116"></a> +<span class="sourceLineNo">117</span> try {<a name="line.117"></a> +<span class="sourceLineNo">118</span> wait(timeout);<a name="line.118"></a> +<span class="sourceLineNo">119</span> } catch (InterruptedException ie) {<a name="line.119"></a> +<span class="sourceLineNo">120</span> // It's ok to ignore<a name="line.120"></a> +<span class="sourceLineNo">121</span> }<a name="line.121"></a> +<span class="sourceLineNo">122</span> break;<a name="line.122"></a> +<span class="sourceLineNo">123</span> }<a name="line.123"></a> +<span class="sourceLineNo">124</span> pool.shutdownNow();<a name="line.124"></a> +<span class="sourceLineNo">125</span> LOG.info("Update chore's pool size from {} to {}", pool.getParallelism(), size);<a name="line.125"></a> +<span class="sourceLineNo">126</span> pool = new ForkJoinPool(size);<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> synchronized void latchCountUp() {<a name="line.129"></a> +<span class="sourceLineNo">130</span> cleanerLatch++;<a name="line.130"></a> +<span class="sourceLineNo">131</span> }<a name="line.131"></a> +<span class="sourceLineNo">132</span><a name="line.132"></a> +<span class="sourceLineNo">133</span> synchronized void latchCountDown() {<a name="line.133"></a> +<span class="sourceLineNo">134</span> cleanerLatch--;<a name="line.134"></a> +<span class="sourceLineNo">135</span> notifyAll();<a name="line.135"></a> +<span class="sourceLineNo">136</span> }<a name="line.136"></a> +<span class="sourceLineNo">137</span><a name="line.137"></a> +<span class="sourceLineNo">138</span> @SuppressWarnings("FutureReturnValueIgnored")<a name="line.138"></a> +<span class="sourceLineNo">139</span> synchronized void submit(ForkJoinTask task) {<a name="line.139"></a> +<span class="sourceLineNo">140</span> pool.submit(task);<a name="line.140"></a> +<span class="sourceLineNo">141</span> }<a name="line.141"></a> +<span class="sourceLineNo">142</span> }<a name="line.142"></a> +<span class="sourceLineNo">143</span> // It may be waste resources for each cleaner chore own its pool,<a name="line.143"></a> +<span class="sourceLineNo">144</span> // so let's make pool for all cleaner chores.<a name="line.144"></a> +<span class="sourceLineNo">145</span> private static volatile DirScanPool POOL;<a name="line.145"></a> +<span class="sourceLineNo">146</span><a name="line.146"></a> +<span class="sourceLineNo">147</span> protected final FileSystem fs;<a name="line.147"></a> +<span class="sourceLineNo">148</span> private final Path oldFileDir;<a name="line.148"></a> +<span class="sourceLineNo">149</span> private final Configuration conf;<a name="line.149"></a> +<span class="sourceLineNo">150</span> protected final Map<String, Object> params;<a name="line.150"></a> +<span class="sourceLineNo">151</span> private final AtomicBoolean enabled = new AtomicBoolean(true);<a name="line.151"></a> +<span class="sourceLineNo">152</span> protected List<T> cleanersChain;<a name="line.152"></a> +<span class="sourceLineNo">153</span><a name="line.153"></a> +<span class="sourceLineNo">154</span> public static void initChorePool(Configuration conf) {<a name="line.154"></a> +<span class="sourceLineNo">155</span> if (POOL == null) {<a name="line.155"></a> +<span class="sourceLineNo">156</span> POOL = new DirScanPool(conf);<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> public CleanerChore(String name, final int sleepPeriod, final Stoppable s, Configuration conf,<a name="line.160"></a> +<span class="sourceLineNo">161</span> FileSystem fs, Path oldFileDir, String confKey) {<a name="line.161"></a> +<span class="sourceLineNo">162</span> this(name, sleepPeriod, s, conf, fs, oldFileDir, confKey, null);<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> /**<a name="line.165"></a> +<span class="sourceLineNo">166</span> * @param name name of the chore being run<a name="line.166"></a> +<span class="sourceLineNo">167</span> * @param sleepPeriod the period of time to sleep between each run<a name="line.167"></a> +<span class="sourceLineNo">168</span> * @param s the stopper<a name="line.168"></a> +<span class="sourceLineNo">169</span> * @param conf configuration to use<a name="line.169"></a> +<span class="sourceLineNo">170</span> * @param fs handle to the FS<a name="line.170"></a> +<span class="sourceLineNo">171</span> * @param oldFileDir the path to the archived files<a name="line.171"></a> +<span class="sourceLineNo">172</span> * @param confKey configuration key for the classes to instantiate<a name="line.172"></a> +<span class="sourceLineNo">173</span> * @param params members could be used in cleaner<a name="line.173"></a> +<span class="sourceLineNo">174</span> */<a name="line.174"></a> +<span class="sourceLineNo">175</span> public CleanerChore(String name, final int sleepPeriod, final Stoppable s, Configuration conf,<a name="line.175"></a> +<span class="sourceLineNo">176</span> FileSystem fs, Path oldFileDir, String confKey, Map<String, Object> params) {<a name="line.176"></a> +<span class="sourceLineNo">177</span> super(name, s, sleepPeriod);<a name="line.177"></a> +<span class="sourceLineNo">178</span><a name="line.178"></a> +<span class="sourceLineNo">179</span> Preconditions.checkNotNull(POOL, "Chore's pool isn't initialized, please call"<a name="line.179"></a> +<span class="sourceLineNo">180</span> + "CleanerChore.initChorePool(Configuration) before new a cleaner chore.");<a name="line.180"></a> +<span class="sourceLineNo">181</span> this.fs = fs;<a name="line.181"></a> +<span class="sourceLineNo">182</span> this.oldFileDir = oldFileDir;<a name="line.182"></a> +<span class="sourceLineNo">183</span> this.conf = conf;<a name="line.183"></a> +<span class="sourceLineNo">184</span> this.params = params;<a name="line.184"></a> +<span class="sourceLineNo">185</span> initCleanerChain(confKey);<a name="line.185"></a> +<span class="sourceLineNo">186</span> }<a name="line.186"></a> +<span class="sourceLineNo">187</span><a name="line.187"></a> +<span class="sourceLineNo">188</span> /**<a name="line.188"></a> +<span class="sourceLineNo">189</span> * Calculate size for cleaner pool.<a name="line.189"></a> +<span class="sourceLineNo">190</span> * @param poolSize size from configuration<a name="line.190"></a> +<span class="sourceLineNo">191</span> * @return size of pool after calculation<a name="line.191"></a> +<span class="sourceLineNo">192</span> */<a name="line.192"></a> +<span class="sourceLineNo">193</span> static int calculatePoolSize(String poolSize) {<a name="line.193"></a> +<span class="sourceLineNo">194</span> if (poolSize.matches("[1-9][0-9]*")) {<a name="line.194"></a> +<span class="sourceLineNo">195</span> // If poolSize is an integer, return it directly,<a name="line.195"></a> +<span class="sourceLineNo">196</span> // but upmost to the number of available processors.<a name="line.196"></a> +<span class="sourceLineNo">197</span> int size = Math.min(Integer.parseInt(poolSize), AVAIL_PROCESSORS);<a name="line.197"></a> +<span class="sourceLineNo">198</span> if (size == AVAIL_PROCESSORS) {<a name="line.198"></a> +<span class="sourceLineNo">199</span> LOG.warn("Use full core processors to scan dir, size={}", size);<a name="line.199"></a> +<span class="sourceLineNo">200</span> }<a name="line.200"></a> +<span class="sourceLineNo">201</span> return size;<a name="line.201"></a> +<span class="sourceLineNo">202</span> } else if (poolSize.matches("0.[0-9]+|1.0")) {<a name="line.202"></a> +<span class="sourceLineNo">203</span> // if poolSize is a double, return poolSize * availableProcessors;<a name="line.203"></a> +<span class="sourceLineNo">204</span> // Ensure that we always return at least one.<a name="line.204"></a> +<span class="sourceLineNo">205</span> int computedThreads = (int) (AVAIL_PROCESSORS * Double.valueOf(poolSize));<a name="line.205"></a> +<span class="sourceLineNo">206</span> if (computedThreads < 1) {<a name="line.206"></a> +<span class="sourceLineNo">207</span> LOG.debug("Computed {} threads for CleanerChore, using 1 instead", computedThreads);<a name="line.207"></a> +<span class="sourceLineNo">208</span> return 1;<a name="line.208"></a> +<span class="sourceLineNo">209</span> }<a name="line.209"></a> +<span class="sourceLineNo">210</span> return computedThreads;<a name="line.210"></a> +<span class="sourceLineNo">211</span> } else {<a name="line.211"></a> +<span class="sourceLineNo">212</span> LOG.error("Unrecognized value: " + poolSize + " for " + CHORE_POOL_SIZE +<a name="line.212"></a> +<span class="sourceLineNo">213</span> ", use default config: " + DEFAULT_CHORE_POOL_SIZE + " instead.");<a name="line.213"></a> +<span class="sourceLineNo">214</span> return calculatePoolSize(DEFAULT_CHORE_POOL_SIZE);<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> /**<a name="line.218"></a> +<span class="sourceLineNo">219</span> * Validate the file to see if it even belongs in the directory. If it is valid, then the file<a name="line.219"></a> +<span class="sourceLineNo">220</span> * will go through the cleaner delegates, but otherwise the file is just deleted.<a name="line.220"></a> +<span class="sourceLineNo">221</span> * @param file full {@link Path} of the file to be checked<a name="line.221"></a> +<span class="sourceLineNo">222</span> * @return <tt>true</tt> if the file is valid, <tt>false</tt> otherwise<a name="line.222"></a> +<span class="sourceLineNo">223</span> */<a name="line.223"></a> +<span class="sourceLineNo">224</span> protected abstract boolean validate(Path file);<a name="line.224"></a> +<span class="sourceLineNo">225</span><a name="line.225"></a> +<span class="sourceLineNo">226</span> /**<a name="line.226"></a> +<span class="sourceLineNo">227</span> * Instantiate and initialize all the file cleaners set in the configuration<a name="line.227"></a> +<span class="sourceLineNo">228</span> * @param confKey key to get the file cleaner classes from the configuration<a name="line.228"></a> +<span class="sourceLineNo">229</span> */<a name="line.229"></a> +<span class="sourceLineNo">230</span> private void initCleanerChain(String confKey) {<a name="line.230"></a> +<span class="sourceLineNo">231</span> this.cleanersChain = new LinkedList<>();<a name="line.231"></a> +<span class="sourceLineNo">232</span> String[] logCleaners = conf.getStrings(confKey);<a name="line.232"></a> +<span class="sourceLineNo">233</span> if (logCleaners != null) {<a name="line.233"></a> +<span class="sourceLineNo">234</span> for (String className : logCleaners) {<a name="line.234"></a> +<span class="sourceLineNo">235</span> T logCleaner = newFileCleaner(className, conf);<a name="line.235"></a> +<span class="sourceLineNo">236</span> if (logCleaner != null) {<a name="line.236"></a> +<span class="sourceLineNo">237</span> LOG.debug("Initialize cleaner={}", className);<a name="line.237"></a> +<span class="sourceLineNo">238</span> this.cleanersChain.add(logCleaner);<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> }<a name="line.241"></a> +<span class="sourceLineNo">242</span> }<a name="line.242"></a> +<span class="sourceLineNo">243</span><a name="line.243"></a> +<span class="sourceLineNo">244</span> @Override<a name="line.244"></a> +<span class="sourceLineNo">245</span> public void onConfigurationChange(Configuration conf) {<a name="line.245"></a> +<span class="sourceLineNo">246</span> if (POOL.canUpdateImmediately(conf)) {<a name="line.246"></a> +<span class="sourceLineNo">247</span> // Can immediately update, no need to wait.<a name="line.247"></a> +<span class="sourceLineNo">248</span> POOL.updatePool(0);<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><a name="line.251"></a> +<span class="sourceLineNo">252</span> /**<a name="line.252"></a> +<span class="sourceLineNo">253</span> * A utility method to create new instances of LogCleanerDelegate based on the class name of the<a name="line.253"></a> +<span class="sourceLineNo">254</span> * LogCleanerDelegate.<a name="line.254"></a> +<span class="sourceLineNo">255</span> * @param className fully qualified class name of the LogCleanerDelegate<a name="line.255"></a> +<span class="sourceLineNo">256</span> * @param conf used configuration<a name="line.256"></a> +<span class="sourceLineNo">257</span> * @return the new instance<a name="line.257"></a> +<span class="sourceLineNo">258</span> */<a name="line.258"></a> +<span class="sourceLineNo">259</span> private T newFileCleaner(String className, Configuration conf) {<a name="line.259"></a> +<span class="sourceLineNo">260</span> try {<a name="line.260"></a> +<span class="sourceLineNo">261</span> Class<? extends FileCleanerDelegate> c = Class.forName(className).asSubclass(<a name="line.261"></a> +<span class="sourceLineNo">262</span> FileCleanerDelegate.class);<a name="line.262"></a> +<span class="sourceLineNo">263</span> @SuppressWarnings("unchecked")<a name="line.263"></a> +<span class="sourceLineNo">264</span> T cleaner = (T) c.getDeclaredConstructor().newInstance();<a name="line.264"></a> +<span class="sourceLineNo">265</span> cleaner.setConf(conf);<a name="line.265"></a> +<span class="sourceLineNo">266</span> cleaner.init(this.params);<a name="line.266"></a> +<span class="sourceLineNo">267</span> return cleaner;<a name="line.267"></a> +<span class="sourceLineNo">268</span> } catch (Exception e) {<a name="line.268"></a> +<span class="sourceLineNo">269</span> LOG.warn("Can NOT create CleanerDelegate={}", className, e);<a name="line.269"></a> +<span class="sourceLineNo">270</span> // skipping if can't instantiate<a name="line.270"></a> +<span class="sourceLineNo">271</span> return null;<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><a name="line.274"></a> +<span class="sourceLineNo">275</span> @Override<a name="line.275"></a> +<span class="sourceLineNo">276</span> protected void chore() {<a name="line.276"></a> +<span class="sourceLineNo">277</span> if (getEnabled()) {<a name="line.277"></a> +<span class="sourceLineNo">278</span> try {<a name="line.278"></a> +<span class="sourceLineNo">279</span> POOL.latchCountUp();<a name="line.279"></a> +<span class="sourceLineNo">280</span> if (runCleaner()) {<a name="line.280"></a> +<span class="sourceLineNo">281</span> LOG.debug("Cleaned all WALs under {}", oldFileDir);<a name="line.281"></a> +<span class="sourceLineNo">282</span> } else {<a name="line.282"></a> +<span class="sourceLineNo">283</span> LOG.warn("WALs outstanding under {}", oldFileDir);<a name="line.283"></a> +<span class="sourceLineNo">284</span> }<a name="line.284"></a> +<span class="sourceLineNo">285</span> } finally {<a name="line.285"></a> +<span class="sourceLineNo">286</span> POOL.latchCountDown();<a name="line.286"></a> +<span class="sourceLineNo">287</span> }<a name="line.287"></a> +<span class="sourceLineNo">288</span> // After each cleaner chore, checks if received reconfigure notification while cleaning.<a name="line.288"></a> +<span class="sourceLineNo">289</span> // First in cleaner turns off notification, to avoid another cleaner updating pool again.<a name="line.289"></a> +<span class="sourceLineNo">290</span> if (POOL.reconfigNotification.compareAndSet(true, false)) {<a name="line.290"></a> +<span class="sourceLineNo">291</span> // This cleaner is waiting for other cleaners finishing their jobs.<a name="line.291"></a> +<span class="sourceLineNo">292</span> // To avoid missing next chore, only wait 0.8 * period, then shutdown.<a name="line.292"></a> +<span class="sourceLineNo">293</span> POOL.updatePool((long) (0.8 * getTimeUnit().toMillis(getPeriod())));<a name="line.293"></a> +<span class="sourceLineNo">294</span> }<a name="line.294"></a> +<span class="sourceLineNo">295</span> } else {<a name="line.295"></a> +<span class="sourceLineNo">296</span> LOG.debug("Cleaner chore disabled! Not cleaning.");<a name="line.296"></a> +<span class="sourceLineNo">297</span> }<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> private void preRunCleaner() {<a name="line.300"></a> +<span class="sourceLineNo">301</span> cleanersChain.forEach(FileCleanerDelegate::preClean);<a name="line.301"></a> +<span class="sourceLineNo">302</span> }<a name="line.302"></a> +<span class="sourceLineNo">303</span><a name="line.303"></a> +<span class="sourceLineNo">304</span> public Boolean runCleaner() {<a name="line.304"></a> +<span class="sourceLineNo">305</span> preRunCleaner();<a name="line.305"></a> +<span class="sourceLineNo">306</span> CleanerTask task = new CleanerTask(this.oldFileDir, true);<a name="line.306"></a> +<span class="sourceLineNo">307</span> POOL.submit(task);<a name="line.307"></a> +<span class="sourceLineNo">308</span> return task.join();<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> * Sort the given list in (descending) order of the space each element takes<a name="line.312"></a> +<span class="sourceLineNo">313</span> * @param dirs the list to sort, element in it should be directory (not file)<a name="line.313"></a> +<span class="sourceLineNo">314</span> */<a name="line.314"></a> +<span class="sourceLineNo">315</span> private void sortByConsumedSpace(List<FileStatus> dirs) {<a name="line.315"></a> +<span class="sourceLineNo">316</span> if (dirs == null || dirs.size() < 2) {<a name="line.316"></a> +<span class="sourceLineNo">317</span> // no need to sort for empty or single directory<a name="line.317"></a> +<span class="sourceLineNo">318</span> return;<a name="line.318"></a> +<span class="sourceLineNo">319</span> }<a name="line.319"></a> +<span class="sourceLineNo">320</span> dirs.sort(new Comparator<FileStatus>() {<a name="line.320"></a> +<span class="sourceLineNo">321</span> HashMap<FileStatus, Long> directorySpaces = new HashMap<>();<a name="line.321"></a> +<span class="sourceLineNo">322</span><a name="line.322"></a> +<span class="sourceLineNo">323</span> @Override<a name="line.323"></a> +<span class="sourceLineNo">324</span> public int compare(FileStatus f1, FileStatus f2) {<a name="line.324"></a> +<span class="sourceLineNo">325</span> long f1ConsumedSpace = getSpace(f1);<a name="line.325"></a> +<span class="sourceLineNo">326</span> long f2ConsumedSpace = getSpace(f2);<a name="line.326"></a> +<span class="sourceLineNo">327</span> return Long.compare(f2ConsumedSpace, f1ConsumedSpace);<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> deletableValidFiles = filteredFiles;<a name="line.330"></a> -<span class="sourceLineNo">331</span> }<a name="line.331"></a> -<span class="sourceLineNo">332</span><a name="line.332"></a> -<span class="sourceLineNo">333</span> Iterable<FileStatus> filesToDelete = Iterables.concat(invalidFiles, deletableValidFiles);<a name="line.333"></a> -<span class="sourceLineNo">334</span> return deleteFiles(filesToDelete) == files.size();<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> * Delete the given files<a name="line.338"></a> -<span class="sourceLineNo">339</span> * @param filesToDelete files to delete<a name="line.339"></a> -<span class="sourceLineNo">340</span> * @return number of deleted files<a name="line.340"></a> -<span class="sourceLineNo">341</span> */<a name="line.341"></a> -<span class="sourceLineNo">342</span> protected int deleteFiles(Iterable<FileStatus> filesToDelete) {<a name="line.342"></a> -<span class="sourceLineNo">343</span> int deletedFileCount = 0;<a name="line.343"></a> -<span class="sourceLineNo">344</span> for (FileStatus file : filesToDelete) {<a name="line.344"></a> -<span class="sourceLineNo">345</span> Path filePath = file.getPath();<a name="line.345"></a> -<span class="sourceLineNo">346</span> LOG.trace("Removing {} from archive", filePath);<a name="line.346"></a> -<span class="sourceLineNo">347</span> try {<a name="line.347"></a> -<span class="sourceLineNo">348</span> boolean success = this.fs.delete(filePath, false);<a name="line.348"></a> -<span class="sourceLineNo">349</span> if (success) {<a name="line.349"></a> -<span class="sourceLineNo">350</span> deletedFileCount++;<a name="line.350"></a> -<span class="sourceLineNo">351</span> } else {<a name="line.351"></a> -<span class="sourceLineNo">352</span> LOG.warn("Attempted to delete:" + filePath<a name="line.352"></a> -<span class="sourceLineNo">353</span> + ", but couldn't. Run cleaner chain and attempt to delete on next pass.");<a name="line.353"></a> -<span class="sourceLineNo">354</span> }<a name="line.354"></a> -<span class="sourceLineNo">355</span> } catch (IOException e) {<a name="line.355"></a> -<span class="sourceLineNo">356</span> e = e instanceof RemoteException ?<a name="line.356"></a> -<span class="sourceLineNo">357</span> ((RemoteException)e).unwrapRemoteException() : e;<a name="line.357"></a> -<span class="sourceLineNo">358</span> LOG.warn("Error while deleting: " + filePath, e);<a name="line.358"></a> -<span class="sourceLineNo">359</span> }<a name="line.359"></a> -<span class="sourceLineNo">360</span> }<a name="line.360"></a> -<span class="sourceLineNo">361</span> return deletedFileCount;<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> @Override<a name="line.364"></a> -<span class="sourceLineNo">365</span> public synchronized void cleanup() {<a name="line.365"></a> -<span class="sourceLineNo">366</span> for (T lc : this.cleanersChain) {<a name="line.366"></a> -<span class="sourceLineNo">367</span> try {<a name="line.367"></a> -<span class="sourceLineNo">368</span> lc.stop("Exiting");<a name="line.368"></a> -<span class="sourceLineNo">369</span> } catch (Throwable t) {<a name="line.369"></a> -<span class="sourceLineNo">370</span> LOG.warn("Stopping", t);<a name="line.370"></a> -<span class="sourceLineNo">371</span> }<a name="line.371"></a> -<span class="sourceLineNo">372</span> }<a name="line.372"></a> -<span class="sourceLineNo">373</span> }<a name="line.373"></a> -<span class="sourceLineNo">374</span><a name="line.374"></a> -<span class="sourceLineNo">375</span> @VisibleForTesting<a name="line.375"></a> -<span class="sourceLineNo">376</span> int getChorePoolSize() {<a name="line.376"></a> -<span class="sourceLineNo">377</span> return CHOREPOOLSIZE;<a name="line.377"></a> -<span class="sourceLineNo">378</span> }<a name="line.378"></a> +<span class="sourceLineNo">330</span> private long getSpace(FileStatus f) {<a name="line.330"></a> +<span class="sourceLineNo">331</span> Long cached = directorySpaces.get(f);<a name="line.331"></a> +<span class="sourceLineNo">332</span> if (cached != null) {<a name="line.332"></a> +<span class="sourceLineNo">333</span> return cached;<a name="line.333"></a> +<span class="sourceLineNo">334</span> }<a name="line.334"></a> +<span class="sourceLineNo">335</span> try {<a name="line.335"></a> +<span class="sourceLineNo">336</span> long space =<a name="line.336"></a> +<span class="sourceLineNo">337</span> f.isDirectory() ? fs.getContentSummary(f.getPath()).getSpaceConsumed() : f.getLen();<a name="line.337"></a> +<span class="sourceLineNo">338</span> directorySpaces.put(f, space);<a name="line.338"></a> +<span class="sourceLineNo">339</span> return space;<a name="line.339"></a> +<span class="sourceLineNo">340</span> } catch (IOException e) {<a name="line.340"></a> +<span class="sourceLineNo">341</span> LOG.trace("Failed to get space consumed by path={}", f, e);<a name="line.341"></a> +<span class="sourceLineNo">342</span> return -1;<a name="line.342"></a> +<span class="sourceLineNo">343</span> }<a name="line.343"></a> +<span class="sourceLineNo">344</span> }<a name="line.344"></a> +<span class="sourceLineNo">345</span> });<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> * Run the given files through each of the cleaners to see if it should be deleted, deleting it if<a name="line.349"></a> +<span class="sourceLineNo">350</span> * necessary.<a name="line.350"></a> +<span class="sourceLineNo">351</span> * @param files List of FileStatus for the files to check (and possibly delete)<a name="line.351"></a> +<span class="sourceLineNo">352</span> * @return true iff successfully deleted all files<a name="line.352"></a> +<span class="sourceLineNo">353</span> */<a name="line.353"></a> +<span class="sourceLineNo">354</span> private boolean checkAndDeleteFiles(List<FileStatus> files) {<a name="line.354"></a> +<span class="sourceLineNo">355</span> if (files == null) {<a name="line.355"></a> +<span class="sourceLineNo">356</span> return true;<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> // first check to see if the path is valid<a name="line.359"></a> +<span class="sourceLineNo">360</span> List<FileStatus> validFiles = Lists.newArrayListWithCapacity(files.size());<a name="line.360"></a> +<span class="sourceLineNo">361</span> List<FileStatus> invalidFiles = Lists.newArrayList();<a name="line.361"></a> +<span class="sourceLineNo">362</span> for (FileStatus file : files) {<a name="line.362"></a> +<span class="sourceLineNo">363</span> if (validate(file.getPath())) {<a name="line.363"></a> +<span class="sourceLineNo">364</span> validFiles.add(file);<a name="line.364"></a> +<span class="sourceLineNo">365</span> } else {<a name="line.365"></a> +<span class="sourceLineNo">366</span> LOG.warn("Found a wrongly formatted file: " + file.getPath() + " - will delete it.");<a name="line.366"></a> +<span class="sourceLineNo">367</span> invalidFiles.add(file);<a name="line.367"></a> +<span class="sourceLineNo">368</span> }<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> Iterable<FileStatus> deletableValidFiles = validFiles;<a name="line.371"></a> +<span class="sourceLineNo">372</span> // check each of the cleaners for the valid files<a name="line.372"></a> +<span class="sourceLineNo">373</span> for (T cleaner : cleanersChain) {<a name="line.373"></a> +<span class="sourceLineNo">374</span> if (cleaner.isStopped() || this.getStopper().isStopped()) {<a name="line.374"></a> +<span class="sourceLineNo">375</span> LOG.warn("A file cleaner" + this.getName() + " is stopped, won't delete any more files in:"<a name="line.375"></a> +<span class="sourceLineNo">376</span> + this.oldFileDir);<a name="line.376"></a> +<span class="sourceLineNo">377</span> return false;<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> /**<a name="line.380"></a> -<span class="sourceLineNo">381</span> * @param enabled<a name="line.381"></a> -<span class="sourceLineNo">382</span> */<a name="line.382"></a> -<span class="sourceLineNo">383</span> public boolean setEnabled(final boolean enabled) {<a name="line.383"></a> -<span class="sourceLineNo">384</span> return this.enabled.getAndSet(enabled);<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> public boolean getEnabled() { return this.enabled.get();<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> private interface Action<T> {<a name="line.390"></a> -<span class="sourceLineNo">391</span> T act() throws IOException;<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> private class CleanerTask extends RecursiveTask<Boolean> {<a name="line.394"></a> -<span class="sourceLineNo">395</span> private final Path dir;<a name="line.395"></a> -<span class="sourceLineNo">396</span> private final boolean root;<a name="line.396"></a> -<span class="sourceLineNo">397</span><a name="line.397"></a> -<span class="sourceLineNo">398</span> CleanerTask(final FileStatus dir, final boolean root) {<a name="line.398"></a> -<span class="sourceLineNo">399</span> this(dir.getPath(), root);<a name="line.399"></a> -<span class="sourceLineNo">400</span> }<a name="line.400"></a> -<span class="sourceLineNo">401</span><a name="line.401"></a> -<span class="sourceLineNo">402</span> CleanerTask(final Path dir, final boolean root) {<a name="line.402"></a> -<span class="sourceLineNo">403</span> this.dir = dir;<a name="line.403"></a> -<span class="sourceLineNo">404</span> this.root = root;<a name="line.404"></a> -<span class="sourceLineNo">405</span> }<a name="line.405"></a> -<span class="sourceLineNo">406</span><a name="line.406"></a> -<span class="sourceLineNo">407</span> @Override<a name="line.407"></a> -<span class="sourceLineNo">408</span> protected Boolean compute() {<a name="line.408"></a> -<span class="sourceLineNo">409</span> LOG.debug("Cleaning under {}", dir);<a name="line.409"></a> -<span class="sourceLineNo">410</span> List<FileStatus> subDirs;<a name="line.410"></a> -<span class="sourceLineNo">411</span> List<FileStatus> files;<a name="line.411"></a> -<span class="sourceLineNo">412</span> try {<a name="line.412"></a> -<span class="sourceLineNo">413</span> subDirs = getFilteredStatus(status -> status.isDirectory());<a name="line.413"></a> -<span class="sourceLineNo">414</span> files = getFilteredStatus(status -> status.isFile());<a name="line.414"></a> -<span class="sourceLineNo">415</span> } catch (IOException ioe) {<a name="line.415"></a> -<span class="sourceLineNo">416</span> LOG.warn(dir + " doesn't exist, just skip it. ", ioe);<a name="line.416"></a> -<span class="sourceLineNo">417</span> return true;<a name="line.417"></a> -<span class="sourceLineNo">418</span> }<a name="line.418"></a> -<span class="sourceLineNo">419</span><a name="line.419"></a> -<span class="sourceLineNo">420</span> boolean nullSubDirs = subDirs == null;<a name="line.420"></a> -<span class="sourceLineNo">421</span> if (nullSubDirs) {<a name="line.421"></a> -<span class="sourceLineNo">422</span> LOG.trace("There is no subdir under {}", dir);<a name="line.422"></a> -<span class="sourceLineNo">423</span> }<a name="line.423"></a> -<span class="sourceLineNo">424</span> if (files == null) {<a name="line.424"></a> -<span class="sourceLineNo">425</span> LOG.trace("There is no file under {}", dir);<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> int capacity = nullSubDirs ? 0 : subDirs.size();<a name="line.428"></a> -<span class="sourceLineNo">429</span> List<CleanerTask> tasks = Lists.newArrayListWithCapacity(capacity);<a name="line.429"></a> -<span class="sourceLineNo">430</span> if (!nullSubDirs) {<a name="line.430"></a> -<span class="sourceLineNo">431</span> sortByConsumedSpace(subDirs);<a name="line.431"></a> -<span class="sourceLineNo">432</span> for (FileStatus subdir : subDirs) {<a name="line.432"></a> -<span class="sourceLineNo">433</span> CleanerTask task = new CleanerTask(subdir, false);<a name="line.433"></a> -<span class="sourceLineNo">434</span> tasks.add(task);<a name="line.434"></a> -<span class="sourceLineNo">435</span> task.fork();<a name="line.435"></a> -<span class="sourceLineNo">436</span> }<a name="line.436"></a> -<span class="sourceLineNo">437</span> }<a name="line.437"></a> -<span class="sourceLineNo">438</span><a name="line.438"></a> -<span class="sourceLineNo">439</span> boolean result = true;<a name="line.439"></a> -<span class="sourceLineNo">440</span> result &= deleteAction(() -> checkAndDeleteFiles(files), "files");<a name="line.440"></a> -<span class="sourceLineNo">441</span> result &= deleteAction(() -> getCleanResult(tasks), "subdirs");<a name="line.441"></a> -<span class="sourceLineNo">442</span> // if and only if files and subdirs under current dir are deleted successfully, and<a name="line.442"></a> -<span class="sourceLineNo">443</span> // it is not the root dir, then task will try to delete it.<a name="line.443"></a> -<span class="sourceLineNo">444</span> if (result && !root) {<a name="line.444"></a> -<span class="sourceLineNo">445</span> result &= deleteAction(() -> fs.delete(dir, false), "dir");<a name="line.445"></a> -<span class="sourceLineNo">446</span> }<a name="line.446"></a> -<span class="sourceLineNo">447</span> return result;<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> /**<a name="line.450"></a> -<span class="sourceLineNo">451</span> * Get FileStatus with filter.<a name="line.451"></a> -<span class="sourceLineNo">452</span> * Pay attention that FSUtils #listStatusWithStatusFilter would return null,<a name="line.452"></a> -<span class="sourceLineNo">453</span> * even though status is empty but not null.<a name="line.453"></a> -<span class="sourceLineNo">454</span> * @param function a filter function<a name="line.454"></a> -<span class="sourceLineNo">455</span> * @return filtered FileStatus<a name="line.455"></a> -<span class="sourceLineNo">456</span> * @throws IOException if there's no such a directory<a name="line.456"></a> -<span class="sourceLineNo">457</span> */<a name="line.457"></a> -<span class="sourceLineNo">458</span> private List<FileStatus> getFilteredStatus(Predicate<FileStatus> function) throws IOException {<a name="line.458"></a> -<span class="sourceLineNo">459</span> return FSUtils.listStatusWithStatusFilter(fs, dir, status -> function.test(status));<a name="line.459"></a> -<span class="sourceLineNo">460</span> }<a name="line.460"></a> -<span class="sourceLineNo">461</span><a name="line.461"></a> -<span class="sourceLineNo">462</span> /**<a name="line.462"></a> -<span class="sourceLineNo">463</span> * Perform a delete on a specified type.<a name="line.463"></a> -<span class="sourceLineNo">464</span> * @param deletion a delete<a name="line.464"></a> -<span class="sourceLineNo">465</span> * @param type possible values are 'files', 'subdirs', 'dirs'<a name="line.465"></a> -<span class="sourceLineNo">466</span> * @return true if it deleted successfully, false otherwise<a name="line.466"></a> -<span class="sourceLineNo">467</span> */<a name="line.467"></a> -<span class="sourceLineNo">468</span> private boolean deleteAction(Action<Boolean> deletion, String type) {<a name="line.468"></a> -<span class="sourceLineNo">469</span> boolean deleted;<a name="line.469"></a> -<span class="sourceLineNo">470</span> try {<a name="line.470"></a> -<span class="sourceLineNo">471</span> LOG.trace("Start deleting {} under {}", type, dir);<a name="line.471"></a> -<span class="sourceLineNo">472</span> deleted = deletion.act();<a name="line.472"></a> -<span class="sourceLineNo">473</span> } catch (IOException ioe) {<a name="line.473"></a> -<span class="sourceLineNo">474</span> LOG.warn("Could not delete {} under {}; {}", type, dir, ioe);<a name="line.474"></a> -<span class="sourceLineNo">475</span> deleted = false;<a name="line.475"></a> -<span class="sourceLineNo">476</span> }<a name="line.476"></a> -<span class="sourceLineNo">477</span> LOG.trace("Finish deleting {} under {}, deleted=", type, dir, deleted);<a name="line.477"></a> -<span class="sourceLineNo">478</span> return deleted;<a name="line.478"></a> -<span class="sourceLineNo">479</span> }<a name="line.479"></a> -<span class="sourceLineNo">480</span><a name="line.480"></a> -<span class="sourceLineNo">481</span> /**<a name="line.481"></a> -<span class="sourceLineNo">482</span> * Get cleaner results of subdirs.<a name="line.482"></a> -<span class="sourceLineNo">483</span> * @param tasks subdirs cleaner tasks<a name="line.483"></a> -<span class="sourceLineNo">484</span> * @return true if all subdirs deleted successfully, false for patial/all failures<a name="line.484"></a> -<span class="sourceLineNo">485</span> * @throws IOException something happen during computation<a name="line.485"></a> -<span class="sourceLineNo">486</span> */<a name="line.486"></a> -<span class="sourceLineNo">487</span> private boolean getCleanResult(List<CleanerTask> tasks) throws IOException {<a name="line.487"></a> -<span class="sourceLineNo">488</span> boolean cleaned = true;<a name="line.488"></a> -<span class="sourceLineNo">489</span> try {<a name="line.489"></a> -<span class="sourceLineNo">490</span> for (CleanerTask task : tasks) {<a name="line.490"></a> -<span class="sourceLineNo">491</span> cleaned &= task.get();<a name="line.491"></a> -<span class="sourceLineNo">492</span> }<a name="line.492"></a> -<span class="sourceLineNo">493</span> } catch (InterruptedException | ExecutionException e) {<a name="line.493"></a> -<span class="sourceLineNo">494</span> throw new IOException(e);<a name="line.494"></a> -<span class="sourceLineNo">495</span> }<a name="line.495"></a> -<span class="sourceLineNo">496</span> return cleaned;<a name="line.496"></a> -<span class="sourceLineNo">497</span> }<a name="line.497"></a> -<span class="sourceLineNo">498</span> }<a name="line.498"></a> -<span class="sourceLineNo">499</span>}<a name="line.499"></a> +<span class="sourceLineNo">380</span> Iterable<FileStatus> filteredFiles = cleaner.getDeletableFiles(deletableValidFiles);<a name="line.380"></a> +<span class="sourceLineNo">381</span><a name="line.381"></a> +<span class="sourceLineNo">382</span> // trace which cleaner is holding on to each file<a name="line.382"></a> +<span class="sourceLineNo">383</span> if (LOG.isTraceEnabled()) {<a name="line.383"></a> +<span class="sourceLineNo">384</span> ImmutableSet<FileStatus> filteredFileSet = ImmutableSet.copyOf(filteredFiles);<a name="line.384"></a> +<span class="sourceLineNo">385</span> for (FileStatus file : deletableValidFiles) {<a name="line.385"></a> +<span class="sourceLineNo">386</span> if (!filteredFileSet.contains(file)) {<a name="line.386"></a> +<span class="sourceLineNo">387</span> LOG.trace(file.getPath() + " is not deletable according to:" + cleaner);<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><a name="line.391"></a> +<span class="sourceLineNo">392</span> deletableValidFiles = filteredFiles;<a name="line.392"></a> +<span class="sourceLineNo">393</span> }<a name="line.393"></a> +<span class="sourceLineNo">394</span><a name="line.394"></a> +<span class="sourceLineNo">395</span> Iterable<FileStatus> filesToDelete = Iterables.concat(invalidFiles, deletableValidFiles);<a name="line.395"></a> +<span class="sourceLineNo">396</span> return deleteFiles(filesToDelete) == files.size();<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> /**<a name="line.399"></a> +<span class="sourceLineNo">400</span> * Delete the given files<a name="line.400"></a> +<span class="sourceLineNo">401</span> * @param filesToDelete files to delete<a name="line.401"></a> +<span class="sourceLineNo">402</span> * @return number of deleted files<a name="line.402"></a> +<span class="sourceLineNo">403</span> */<a name="line.403"></a> +<span class="sourceLineNo">404</span> protected int deleteFiles(Iterable<FileStatus> filesToDelete) {<a name="line.404"></a> +<span class="sourceLineNo">405</span> int deletedFileCount = 0;<a name="line.405"></a> +<span class="sourceLineNo">406</span> for (FileStatus file : filesToDelete) {<a name="line.406"></a> +<span class="sourceLineNo">407</span> Path filePath = file.getPath();<a name="line.407"></a> +<span class="sourceLineNo">408</span> LOG.trace("Removing {} from archive", filePath);<a name="line.408"></a> +<span class="sourceLineNo">409</span> try {<a name="line.409"></a> +<span class="sourceLineNo">410</span> boolean success = this.fs.delete(filePath, false);<a name="line.410"></a> +<span class="sourceLineNo">411</span> if (success) {<a name="line.411"></a> +<span class="sourceLineNo">412</span> deletedFileCount++;<a name="line.412"></a> +<span class="sourceLineNo">413</span> } else {<a name="line.413"></a> +<span class="sourceLineNo">414</span> LOG.warn("Attempted to delete:" + filePath<a name="line.414"></a> +<span class="sourceLineNo">415</span> + ", but couldn't. Run cleaner chain and attempt to delete on next pass.");<a name="line.415"></a> +<span class="sourceLineNo">416</span> }<a name="line.416"></a> +<span class="sourceLineNo">417</span> } catch (IOException e) {<a name="line.417"></a> +<span class="sourceLineNo">418</span> e = e instanceof RemoteException ?<a name="line.418"></a> +<span class="sourceLineNo">419</span> ((RemoteException)e).unwrapRemoteException() : e;<a name="line.419"></a> +<span class="sourceLineNo">420</span> LOG.warn("Error while deleting: " + filePath, e);<a name="line.420"></a> +<span class="sourceLineNo">421</span> }<a name="line.421"></a> +<span class="sourceLineNo">422</span> }<a name="line.422"></a> +<span class="sourceLineNo">423</span> return deletedFileCount;<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> @Override<a name="line.426"></a> +<span class="sourceLineNo">427</span> public synchronized void cleanup() {<a name="line.427"></a> +<span class="sourceLineNo">428</span> for (T lc : this.cleanersChain) {<a name="line.428"></a> +<span class="sourceLineNo">429</span> try {<a name="line.429"></a> +<span class="sourceLineNo">430</span> lc.stop("Exiting");<a name="line.430"></a> +<span class="sourceLineNo">431</span> } catch (Throwable t) {<a name="line.431"></a> +<span class="sourceLineNo">432</span> LOG.warn("Stopping", t);<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><a name="line.436"></a> +<span class="sourceLineNo">437</span> @VisibleForTesting<a name="line.437"></a> +<span class="sourceLineNo">438</span> int getChorePoolSize() {<a name="line.438"></a> +<span class="sourceLineNo">439</span> return POOL.size;<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> * @param enabled<a name="line.443"></a> +<span class="sourceLineNo">444</span> */<a name="line.444"></a> +<span class="sourceLineNo">445</span> public boolean setEnabled(final boolean enabled) {<a name="line.445"></a> +<span class="sourceLineNo">446</span> return this.enabled.getAndSet(enabled);<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> public boolean getEnabled() { return this.enabled.get();<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> private interface Action<T> {<a name="line.452"></a> +<span class="sourceLineNo">453</span> T act() throws IOException;<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 class CleanerTask extends RecursiveTask<Boolean> {<a name="line.456"></a> +<span class="sourceLineNo">457</span> private final Path dir;<a name="line.457"></a> +<span class="sourceLineNo">458</span> private final boolean root;<a name="line.458"></a> +<span class="sourceLineNo">459</span><a name="line.459"></a> +<span class="sourceLineNo">460</span> CleanerTask(final FileStatus dir, final boolean root) {<a name="line.460"></a> +<span class="sourceLineNo">461</span> this(dir.getPath(), root);<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> CleanerTask(final Path dir, final boolean root) {<a name="line.464"></a> +<span class="sourceLineNo">465</span> this.dir = dir;<a name="line.465"></a> +<span class="sourceLineNo">466</span> this.root = root;<a name="line.466"></a> +<span class="sourceLineNo">467</span> }<a name="line.467"></a> +<span class="sourceLineNo">468</span><a name="line.468"></a> +<span class="sourceLineNo">469</span> @Override<a name="line.469"></a> +<span class="sourceLineNo">470</span> protected Boolean compute() {<a name="line.470"></a> +<span class="sourceLineNo">471</span> LOG.debug("Cleaning under {}", dir);<a name="line.471"></a> +<span class="sourceLineNo">472</span> List<FileStatus> subDirs;<a name="line.472"></a> +<span class="sourceLineNo">473</span> List<FileStatus> files;<a name="line.473"></a> +<span class="sourceLineNo">474</span> try {<a name="line.474"></a> +<span class="sourceLineNo">475</span> subDirs = <TRUNCATED>
