http://git-wip-us.apache.org/repos/asf/hbase-site/blob/ce958bce/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableRecordReaderImpl.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableRecordReaderImpl.html b/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableRecordReaderImpl.html index 288dfb3..6df12f7 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableRecordReaderImpl.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableRecordReaderImpl.html @@ -162,109 +162,111 @@ <span class="sourceLineNo">154</span> }<a name="line.154"></a> <span class="sourceLineNo">155</span><a name="line.155"></a> <span class="sourceLineNo">156</span> public void close() {<a name="line.156"></a> -<span class="sourceLineNo">157</span> this.scanner.close();<a name="line.157"></a> -<span class="sourceLineNo">158</span> try {<a name="line.158"></a> -<span class="sourceLineNo">159</span> this.htable.close();<a name="line.159"></a> -<span class="sourceLineNo">160</span> } catch (IOException ioe) {<a name="line.160"></a> -<span class="sourceLineNo">161</span> LOG.warn("Error closing table", ioe);<a name="line.161"></a> -<span class="sourceLineNo">162</span> }<a name="line.162"></a> -<span class="sourceLineNo">163</span> }<a name="line.163"></a> -<span class="sourceLineNo">164</span><a name="line.164"></a> -<span class="sourceLineNo">165</span> /**<a name="line.165"></a> -<span class="sourceLineNo">166</span> * @return ImmutableBytesWritable<a name="line.166"></a> -<span class="sourceLineNo">167</span> *<a name="line.167"></a> -<span class="sourceLineNo">168</span> * @see org.apache.hadoop.mapred.RecordReader#createKey()<a name="line.168"></a> -<span class="sourceLineNo">169</span> */<a name="line.169"></a> -<span class="sourceLineNo">170</span> public ImmutableBytesWritable createKey() {<a name="line.170"></a> -<span class="sourceLineNo">171</span> return new ImmutableBytesWritable();<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> * @return RowResult<a name="line.175"></a> -<span class="sourceLineNo">176</span> *<a name="line.176"></a> -<span class="sourceLineNo">177</span> * @see org.apache.hadoop.mapred.RecordReader#createValue()<a name="line.177"></a> -<span class="sourceLineNo">178</span> */<a name="line.178"></a> -<span class="sourceLineNo">179</span> public Result createValue() {<a name="line.179"></a> -<span class="sourceLineNo">180</span> return new Result();<a name="line.180"></a> -<span class="sourceLineNo">181</span> }<a name="line.181"></a> -<span class="sourceLineNo">182</span><a name="line.182"></a> -<span class="sourceLineNo">183</span> public long getPos() {<a name="line.183"></a> -<span class="sourceLineNo">184</span> // This should be the ordinal tuple in the range;<a name="line.184"></a> -<span class="sourceLineNo">185</span> // not clear how to calculate...<a name="line.185"></a> -<span class="sourceLineNo">186</span> return 0;<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> public float getProgress() {<a name="line.189"></a> -<span class="sourceLineNo">190</span> // Depends on the total number of tuples and getPos<a name="line.190"></a> -<span class="sourceLineNo">191</span> return 0;<a name="line.191"></a> -<span class="sourceLineNo">192</span> }<a name="line.192"></a> -<span class="sourceLineNo">193</span><a name="line.193"></a> -<span class="sourceLineNo">194</span> /**<a name="line.194"></a> -<span class="sourceLineNo">195</span> * @param key HStoreKey as input key.<a name="line.195"></a> -<span class="sourceLineNo">196</span> * @param value MapWritable as input value<a name="line.196"></a> -<span class="sourceLineNo">197</span> * @return true if there was more data<a name="line.197"></a> -<span class="sourceLineNo">198</span> * @throws IOException<a name="line.198"></a> -<span class="sourceLineNo">199</span> */<a name="line.199"></a> -<span class="sourceLineNo">200</span> public boolean next(ImmutableBytesWritable key, Result value)<a name="line.200"></a> -<span class="sourceLineNo">201</span> throws IOException {<a name="line.201"></a> -<span class="sourceLineNo">202</span> Result result;<a name="line.202"></a> -<span class="sourceLineNo">203</span> try {<a name="line.203"></a> -<span class="sourceLineNo">204</span> try {<a name="line.204"></a> -<span class="sourceLineNo">205</span> result = this.scanner.next();<a name="line.205"></a> -<span class="sourceLineNo">206</span> if (logScannerActivity) {<a name="line.206"></a> -<span class="sourceLineNo">207</span> rowcount ++;<a name="line.207"></a> -<span class="sourceLineNo">208</span> if (rowcount >= logPerRowCount) {<a name="line.208"></a> -<span class="sourceLineNo">209</span> long now = System.currentTimeMillis();<a name="line.209"></a> -<span class="sourceLineNo">210</span> LOG.info("Mapper took " + (now-timestamp)<a name="line.210"></a> -<span class="sourceLineNo">211</span> + "ms to process " + rowcount + " rows");<a name="line.211"></a> -<span class="sourceLineNo">212</span> timestamp = now;<a name="line.212"></a> -<span class="sourceLineNo">213</span> rowcount = 0;<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> } catch (IOException e) {<a name="line.216"></a> -<span class="sourceLineNo">217</span> // do not retry if the exception tells us not to do so<a name="line.217"></a> -<span class="sourceLineNo">218</span> if (e instanceof DoNotRetryIOException) {<a name="line.218"></a> -<span class="sourceLineNo">219</span> throw e;<a name="line.219"></a> -<span class="sourceLineNo">220</span> }<a name="line.220"></a> -<span class="sourceLineNo">221</span> // try to handle all other IOExceptions by restarting<a name="line.221"></a> -<span class="sourceLineNo">222</span> // the scanner, if the second call fails, it will be rethrown<a name="line.222"></a> -<span class="sourceLineNo">223</span> LOG.debug("recovered from " + StringUtils.stringifyException(e));<a name="line.223"></a> -<span class="sourceLineNo">224</span> if (lastSuccessfulRow == null) {<a name="line.224"></a> -<span class="sourceLineNo">225</span> LOG.warn("We are restarting the first next() invocation," +<a name="line.225"></a> -<span class="sourceLineNo">226</span> " if your mapper has restarted a few other times like this" +<a name="line.226"></a> -<span class="sourceLineNo">227</span> " then you should consider killing this job and investigate" +<a name="line.227"></a> -<span class="sourceLineNo">228</span> " why it's taking so long.");<a name="line.228"></a> -<span class="sourceLineNo">229</span> }<a name="line.229"></a> -<span class="sourceLineNo">230</span> if (lastSuccessfulRow == null) {<a name="line.230"></a> -<span class="sourceLineNo">231</span> restart(startRow);<a name="line.231"></a> -<span class="sourceLineNo">232</span> } else {<a name="line.232"></a> -<span class="sourceLineNo">233</span> restart(lastSuccessfulRow);<a name="line.233"></a> -<span class="sourceLineNo">234</span> this.scanner.next(); // skip presumed already mapped row<a name="line.234"></a> -<span class="sourceLineNo">235</span> }<a name="line.235"></a> -<span class="sourceLineNo">236</span> result = this.scanner.next();<a name="line.236"></a> -<span class="sourceLineNo">237</span> }<a name="line.237"></a> -<span class="sourceLineNo">238</span><a name="line.238"></a> -<span class="sourceLineNo">239</span> if (result != null && result.size() > 0) {<a name="line.239"></a> -<span class="sourceLineNo">240</span> key.set(result.getRow());<a name="line.240"></a> -<span class="sourceLineNo">241</span> lastSuccessfulRow = key.get();<a name="line.241"></a> -<span class="sourceLineNo">242</span> value.copyFrom(result);<a name="line.242"></a> -<span class="sourceLineNo">243</span> return true;<a name="line.243"></a> -<span class="sourceLineNo">244</span> }<a name="line.244"></a> -<span class="sourceLineNo">245</span> return false;<a name="line.245"></a> -<span class="sourceLineNo">246</span> } catch (IOException ioe) {<a name="line.246"></a> -<span class="sourceLineNo">247</span> if (logScannerActivity) {<a name="line.247"></a> -<span class="sourceLineNo">248</span> long now = System.currentTimeMillis();<a name="line.248"></a> -<span class="sourceLineNo">249</span> LOG.info("Mapper took " + (now-timestamp)<a name="line.249"></a> -<span class="sourceLineNo">250</span> + "ms to process " + rowcount + " rows");<a name="line.250"></a> -<span class="sourceLineNo">251</span> LOG.info(ioe);<a name="line.251"></a> -<span class="sourceLineNo">252</span> String lastRow = lastSuccessfulRow == null ?<a name="line.252"></a> -<span class="sourceLineNo">253</span> "null" : Bytes.toStringBinary(lastSuccessfulRow);<a name="line.253"></a> -<span class="sourceLineNo">254</span> LOG.info("lastSuccessfulRow=" + lastRow);<a name="line.254"></a> -<span class="sourceLineNo">255</span> }<a name="line.255"></a> -<span class="sourceLineNo">256</span> throw ioe;<a name="line.256"></a> -<span class="sourceLineNo">257</span> }<a name="line.257"></a> -<span class="sourceLineNo">258</span> }<a name="line.258"></a> -<span class="sourceLineNo">259</span>}<a name="line.259"></a> +<span class="sourceLineNo">157</span> if (this.scanner != null) {<a name="line.157"></a> +<span class="sourceLineNo">158</span> this.scanner.close();<a name="line.158"></a> +<span class="sourceLineNo">159</span> }<a name="line.159"></a> +<span class="sourceLineNo">160</span> try {<a name="line.160"></a> +<span class="sourceLineNo">161</span> this.htable.close();<a name="line.161"></a> +<span class="sourceLineNo">162</span> } catch (IOException ioe) {<a name="line.162"></a> +<span class="sourceLineNo">163</span> LOG.warn("Error closing table", ioe);<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><a name="line.166"></a> +<span class="sourceLineNo">167</span> /**<a name="line.167"></a> +<span class="sourceLineNo">168</span> * @return ImmutableBytesWritable<a name="line.168"></a> +<span class="sourceLineNo">169</span> *<a name="line.169"></a> +<span class="sourceLineNo">170</span> * @see org.apache.hadoop.mapred.RecordReader#createKey()<a name="line.170"></a> +<span class="sourceLineNo">171</span> */<a name="line.171"></a> +<span class="sourceLineNo">172</span> public ImmutableBytesWritable createKey() {<a name="line.172"></a> +<span class="sourceLineNo">173</span> return new ImmutableBytesWritable();<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> /**<a name="line.176"></a> +<span class="sourceLineNo">177</span> * @return RowResult<a name="line.177"></a> +<span class="sourceLineNo">178</span> *<a name="line.178"></a> +<span class="sourceLineNo">179</span> * @see org.apache.hadoop.mapred.RecordReader#createValue()<a name="line.179"></a> +<span class="sourceLineNo">180</span> */<a name="line.180"></a> +<span class="sourceLineNo">181</span> public Result createValue() {<a name="line.181"></a> +<span class="sourceLineNo">182</span> return new Result();<a name="line.182"></a> +<span class="sourceLineNo">183</span> }<a name="line.183"></a> +<span class="sourceLineNo">184</span><a name="line.184"></a> +<span class="sourceLineNo">185</span> public long getPos() {<a name="line.185"></a> +<span class="sourceLineNo">186</span> // This should be the ordinal tuple in the range;<a name="line.186"></a> +<span class="sourceLineNo">187</span> // not clear how to calculate...<a name="line.187"></a> +<span class="sourceLineNo">188</span> return 0;<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> public float getProgress() {<a name="line.191"></a> +<span class="sourceLineNo">192</span> // Depends on the total number of tuples and getPos<a name="line.192"></a> +<span class="sourceLineNo">193</span> return 0;<a name="line.193"></a> +<span class="sourceLineNo">194</span> }<a name="line.194"></a> +<span class="sourceLineNo">195</span><a name="line.195"></a> +<span class="sourceLineNo">196</span> /**<a name="line.196"></a> +<span class="sourceLineNo">197</span> * @param key HStoreKey as input key.<a name="line.197"></a> +<span class="sourceLineNo">198</span> * @param value MapWritable as input value<a name="line.198"></a> +<span class="sourceLineNo">199</span> * @return true if there was more data<a name="line.199"></a> +<span class="sourceLineNo">200</span> * @throws IOException<a name="line.200"></a> +<span class="sourceLineNo">201</span> */<a name="line.201"></a> +<span class="sourceLineNo">202</span> public boolean next(ImmutableBytesWritable key, Result value)<a name="line.202"></a> +<span class="sourceLineNo">203</span> throws IOException {<a name="line.203"></a> +<span class="sourceLineNo">204</span> Result result;<a name="line.204"></a> +<span class="sourceLineNo">205</span> try {<a name="line.205"></a> +<span class="sourceLineNo">206</span> try {<a name="line.206"></a> +<span class="sourceLineNo">207</span> result = this.scanner.next();<a name="line.207"></a> +<span class="sourceLineNo">208</span> if (logScannerActivity) {<a name="line.208"></a> +<span class="sourceLineNo">209</span> rowcount ++;<a name="line.209"></a> +<span class="sourceLineNo">210</span> if (rowcount >= logPerRowCount) {<a name="line.210"></a> +<span class="sourceLineNo">211</span> long now = System.currentTimeMillis();<a name="line.211"></a> +<span class="sourceLineNo">212</span> LOG.info("Mapper took " + (now-timestamp)<a name="line.212"></a> +<span class="sourceLineNo">213</span> + "ms to process " + rowcount + " rows");<a name="line.213"></a> +<span class="sourceLineNo">214</span> timestamp = now;<a name="line.214"></a> +<span class="sourceLineNo">215</span> rowcount = 0;<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> } catch (IOException e) {<a name="line.218"></a> +<span class="sourceLineNo">219</span> // do not retry if the exception tells us not to do so<a name="line.219"></a> +<span class="sourceLineNo">220</span> if (e instanceof DoNotRetryIOException) {<a name="line.220"></a> +<span class="sourceLineNo">221</span> throw e;<a name="line.221"></a> +<span class="sourceLineNo">222</span> }<a name="line.222"></a> +<span class="sourceLineNo">223</span> // try to handle all other IOExceptions by restarting<a name="line.223"></a> +<span class="sourceLineNo">224</span> // the scanner, if the second call fails, it will be rethrown<a name="line.224"></a> +<span class="sourceLineNo">225</span> LOG.debug("recovered from " + StringUtils.stringifyException(e));<a name="line.225"></a> +<span class="sourceLineNo">226</span> if (lastSuccessfulRow == null) {<a name="line.226"></a> +<span class="sourceLineNo">227</span> LOG.warn("We are restarting the first next() invocation," +<a name="line.227"></a> +<span class="sourceLineNo">228</span> " if your mapper has restarted a few other times like this" +<a name="line.228"></a> +<span class="sourceLineNo">229</span> " then you should consider killing this job and investigate" +<a name="line.229"></a> +<span class="sourceLineNo">230</span> " why it's taking so long.");<a name="line.230"></a> +<span class="sourceLineNo">231</span> }<a name="line.231"></a> +<span class="sourceLineNo">232</span> if (lastSuccessfulRow == null) {<a name="line.232"></a> +<span class="sourceLineNo">233</span> restart(startRow);<a name="line.233"></a> +<span class="sourceLineNo">234</span> } else {<a name="line.234"></a> +<span class="sourceLineNo">235</span> restart(lastSuccessfulRow);<a name="line.235"></a> +<span class="sourceLineNo">236</span> this.scanner.next(); // skip presumed already mapped row<a name="line.236"></a> +<span class="sourceLineNo">237</span> }<a name="line.237"></a> +<span class="sourceLineNo">238</span> result = this.scanner.next();<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> if (result != null && result.size() > 0) {<a name="line.241"></a> +<span class="sourceLineNo">242</span> key.set(result.getRow());<a name="line.242"></a> +<span class="sourceLineNo">243</span> lastSuccessfulRow = key.get();<a name="line.243"></a> +<span class="sourceLineNo">244</span> value.copyFrom(result);<a name="line.244"></a> +<span class="sourceLineNo">245</span> return true;<a name="line.245"></a> +<span class="sourceLineNo">246</span> }<a name="line.246"></a> +<span class="sourceLineNo">247</span> return false;<a name="line.247"></a> +<span class="sourceLineNo">248</span> } catch (IOException ioe) {<a name="line.248"></a> +<span class="sourceLineNo">249</span> if (logScannerActivity) {<a name="line.249"></a> +<span class="sourceLineNo">250</span> long now = System.currentTimeMillis();<a name="line.250"></a> +<span class="sourceLineNo">251</span> LOG.info("Mapper took " + (now-timestamp)<a name="line.251"></a> +<span class="sourceLineNo">252</span> + "ms to process " + rowcount + " rows");<a name="line.252"></a> +<span class="sourceLineNo">253</span> LOG.info(ioe);<a name="line.253"></a> +<span class="sourceLineNo">254</span> String lastRow = lastSuccessfulRow == null ?<a name="line.254"></a> +<span class="sourceLineNo">255</span> "null" : Bytes.toStringBinary(lastSuccessfulRow);<a name="line.255"></a> +<span class="sourceLineNo">256</span> LOG.info("lastSuccessfulRow=" + lastRow);<a name="line.256"></a> +<span class="sourceLineNo">257</span> }<a name="line.257"></a> +<span class="sourceLineNo">258</span> throw ioe;<a name="line.258"></a> +<span class="sourceLineNo">259</span> }<a name="line.259"></a> +<span class="sourceLineNo">260</span> }<a name="line.260"></a> +<span class="sourceLineNo">261</span>}<a name="line.261"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/ce958bce/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html index a10017d..099a926 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.html @@ -796,242 +796,243 @@ <span class="sourceLineNo">788</span> org.apache.hadoop.hbase.client.Put.class, // hbase-client<a name="line.788"></a> <span class="sourceLineNo">789</span> org.apache.hadoop.hbase.CompatibilityFactory.class, // hbase-hadoop-compat<a name="line.789"></a> <span class="sourceLineNo">790</span> org.apache.hadoop.hbase.mapreduce.TableMapper.class, // hbase-server<a name="line.790"></a> -<span class="sourceLineNo">791</span> prefixTreeCodecClass, // hbase-prefix-tree (if null will be skipped)<a name="line.791"></a> -<span class="sourceLineNo">792</span> // pull necessary dependencies<a name="line.792"></a> -<span class="sourceLineNo">793</span> org.apache.zookeeper.ZooKeeper.class,<a name="line.793"></a> -<span class="sourceLineNo">794</span> io.netty.channel.Channel.class,<a name="line.794"></a> -<span class="sourceLineNo">795</span> com.google.protobuf.Message.class,<a name="line.795"></a> -<span class="sourceLineNo">796</span> com.google.common.collect.Lists.class,<a name="line.796"></a> -<span class="sourceLineNo">797</span> org.apache.htrace.Trace.class,<a name="line.797"></a> -<span class="sourceLineNo">798</span> com.codahale.metrics.MetricRegistry.class);<a name="line.798"></a> -<span class="sourceLineNo">799</span> }<a name="line.799"></a> -<span class="sourceLineNo">800</span><a name="line.800"></a> -<span class="sourceLineNo">801</span> /**<a name="line.801"></a> -<span class="sourceLineNo">802</span> * Returns a classpath string built from the content of the "tmpjars" value in {@code conf}.<a name="line.802"></a> -<span class="sourceLineNo">803</span> * Also exposed to shell scripts via `bin/hbase mapredcp`.<a name="line.803"></a> -<span class="sourceLineNo">804</span> */<a name="line.804"></a> -<span class="sourceLineNo">805</span> public static String buildDependencyClasspath(Configuration conf) {<a name="line.805"></a> -<span class="sourceLineNo">806</span> if (conf == null) {<a name="line.806"></a> -<span class="sourceLineNo">807</span> throw new IllegalArgumentException("Must provide a configuration object.");<a name="line.807"></a> -<span class="sourceLineNo">808</span> }<a name="line.808"></a> -<span class="sourceLineNo">809</span> Set<String> paths = new HashSet<String>(conf.getStringCollection("tmpjars"));<a name="line.809"></a> -<span class="sourceLineNo">810</span> if (paths.isEmpty()) {<a name="line.810"></a> -<span class="sourceLineNo">811</span> throw new IllegalArgumentException("Configuration contains no tmpjars.");<a name="line.811"></a> -<span class="sourceLineNo">812</span> }<a name="line.812"></a> -<span class="sourceLineNo">813</span> StringBuilder sb = new StringBuilder();<a name="line.813"></a> -<span class="sourceLineNo">814</span> for (String s : paths) {<a name="line.814"></a> -<span class="sourceLineNo">815</span> // entries can take the form 'file:/path/to/file.jar'.<a name="line.815"></a> -<span class="sourceLineNo">816</span> int idx = s.indexOf(":");<a name="line.816"></a> -<span class="sourceLineNo">817</span> if (idx != -1) s = s.substring(idx + 1);<a name="line.817"></a> -<span class="sourceLineNo">818</span> if (sb.length() > 0) sb.append(File.pathSeparator);<a name="line.818"></a> -<span class="sourceLineNo">819</span> sb.append(s);<a name="line.819"></a> -<span class="sourceLineNo">820</span> }<a name="line.820"></a> -<span class="sourceLineNo">821</span> return sb.toString();<a name="line.821"></a> -<span class="sourceLineNo">822</span> }<a name="line.822"></a> -<span class="sourceLineNo">823</span><a name="line.823"></a> -<span class="sourceLineNo">824</span> /**<a name="line.824"></a> -<span class="sourceLineNo">825</span> * Add the HBase dependency jars as well as jars for any of the configured<a name="line.825"></a> -<span class="sourceLineNo">826</span> * job classes to the job configuration, so that JobClient will ship them<a name="line.826"></a> -<span class="sourceLineNo">827</span> * to the cluster and add them to the DistributedCache.<a name="line.827"></a> -<span class="sourceLineNo">828</span> */<a name="line.828"></a> -<span class="sourceLineNo">829</span> public static void addDependencyJars(Job job) throws IOException {<a name="line.829"></a> -<span class="sourceLineNo">830</span> addHBaseDependencyJars(job.getConfiguration());<a name="line.830"></a> -<span class="sourceLineNo">831</span> try {<a name="line.831"></a> -<span class="sourceLineNo">832</span> addDependencyJarsForClasses(job.getConfiguration(),<a name="line.832"></a> -<span class="sourceLineNo">833</span> // when making changes here, consider also mapred.TableMapReduceUtil<a name="line.833"></a> -<span class="sourceLineNo">834</span> // pull job classes<a name="line.834"></a> -<span class="sourceLineNo">835</span> job.getMapOutputKeyClass(),<a name="line.835"></a> -<span class="sourceLineNo">836</span> job.getMapOutputValueClass(),<a name="line.836"></a> -<span class="sourceLineNo">837</span> job.getInputFormatClass(),<a name="line.837"></a> -<span class="sourceLineNo">838</span> job.getOutputKeyClass(),<a name="line.838"></a> -<span class="sourceLineNo">839</span> job.getOutputValueClass(),<a name="line.839"></a> -<span class="sourceLineNo">840</span> job.getOutputFormatClass(),<a name="line.840"></a> -<span class="sourceLineNo">841</span> job.getPartitionerClass(),<a name="line.841"></a> -<span class="sourceLineNo">842</span> job.getCombinerClass());<a name="line.842"></a> -<span class="sourceLineNo">843</span> } catch (ClassNotFoundException e) {<a name="line.843"></a> -<span class="sourceLineNo">844</span> throw new IOException(e);<a name="line.844"></a> -<span class="sourceLineNo">845</span> }<a name="line.845"></a> -<span class="sourceLineNo">846</span> }<a name="line.846"></a> -<span class="sourceLineNo">847</span><a name="line.847"></a> -<span class="sourceLineNo">848</span> /**<a name="line.848"></a> -<span class="sourceLineNo">849</span> * Add the jars containing the given classes to the job's configuration<a name="line.849"></a> -<span class="sourceLineNo">850</span> * such that JobClient will ship them to the cluster and add them to<a name="line.850"></a> -<span class="sourceLineNo">851</span> * the DistributedCache.<a name="line.851"></a> -<span class="sourceLineNo">852</span> * @deprecated rely on {@link #addDependencyJars(Job)} instead.<a name="line.852"></a> -<span class="sourceLineNo">853</span> */<a name="line.853"></a> -<span class="sourceLineNo">854</span> @Deprecated<a name="line.854"></a> -<span class="sourceLineNo">855</span> public static void addDependencyJars(Configuration conf,<a name="line.855"></a> -<span class="sourceLineNo">856</span> Class<?>... classes) throws IOException {<a name="line.856"></a> -<span class="sourceLineNo">857</span> LOG.warn("The addDependencyJars(Configuration, Class<?>...) method has been deprecated since it"<a name="line.857"></a> -<span class="sourceLineNo">858</span> + " is easy to use incorrectly. Most users should rely on addDependencyJars(Job) " +<a name="line.858"></a> -<span class="sourceLineNo">859</span> "instead. See HBASE-8386 for more details.");<a name="line.859"></a> -<span class="sourceLineNo">860</span> addDependencyJarsForClasses(conf, classes);<a name="line.860"></a> -<span class="sourceLineNo">861</span> }<a name="line.861"></a> -<span class="sourceLineNo">862</span><a name="line.862"></a> -<span class="sourceLineNo">863</span> /**<a name="line.863"></a> -<span class="sourceLineNo">864</span> * Add the jars containing the given classes to the job's configuration<a name="line.864"></a> -<span class="sourceLineNo">865</span> * such that JobClient will ship them to the cluster and add them to<a name="line.865"></a> -<span class="sourceLineNo">866</span> * the DistributedCache.<a name="line.866"></a> -<span class="sourceLineNo">867</span> *<a name="line.867"></a> -<span class="sourceLineNo">868</span> * N.B. that this method at most adds one jar per class given. If there is more than one<a name="line.868"></a> -<span class="sourceLineNo">869</span> * jar available containing a class with the same name as a given class, we don't define<a name="line.869"></a> -<span class="sourceLineNo">870</span> * which of those jars might be chosen.<a name="line.870"></a> -<span class="sourceLineNo">871</span> *<a name="line.871"></a> -<span class="sourceLineNo">872</span> * @param conf The Hadoop Configuration to modify<a name="line.872"></a> -<span class="sourceLineNo">873</span> * @param classes will add just those dependencies needed to find the given classes<a name="line.873"></a> -<span class="sourceLineNo">874</span> * @throws IOException if an underlying library call fails.<a name="line.874"></a> -<span class="sourceLineNo">875</span> */<a name="line.875"></a> -<span class="sourceLineNo">876</span> @InterfaceAudience.Private<a name="line.876"></a> -<span class="sourceLineNo">877</span> public static void addDependencyJarsForClasses(Configuration conf,<a name="line.877"></a> -<span class="sourceLineNo">878</span> Class<?>... classes) throws IOException {<a name="line.878"></a> -<span class="sourceLineNo">879</span><a name="line.879"></a> -<span class="sourceLineNo">880</span> FileSystem localFs = FileSystem.getLocal(conf);<a name="line.880"></a> -<span class="sourceLineNo">881</span> Set<String> jars = new HashSet<String>();<a name="line.881"></a> -<span class="sourceLineNo">882</span> // Add jars that are already in the tmpjars variable<a name="line.882"></a> -<span class="sourceLineNo">883</span> jars.addAll(conf.getStringCollection("tmpjars"));<a name="line.883"></a> -<span class="sourceLineNo">884</span><a name="line.884"></a> -<span class="sourceLineNo">885</span> // add jars as we find them to a map of contents jar name so that we can avoid<a name="line.885"></a> -<span class="sourceLineNo">886</span> // creating new jars for classes that have already been packaged.<a name="line.886"></a> -<span class="sourceLineNo">887</span> Map<String, String> packagedClasses = new HashMap<String, String>();<a name="line.887"></a> -<span class="sourceLineNo">888</span><a name="line.888"></a> -<span class="sourceLineNo">889</span> // Add jars containing the specified classes<a name="line.889"></a> -<span class="sourceLineNo">890</span> for (Class<?> clazz : classes) {<a name="line.890"></a> -<span class="sourceLineNo">891</span> if (clazz == null) continue;<a name="line.891"></a> -<span class="sourceLineNo">892</span><a name="line.892"></a> -<span class="sourceLineNo">893</span> Path path = findOrCreateJar(clazz, localFs, packagedClasses);<a name="line.893"></a> -<span class="sourceLineNo">894</span> if (path == null) {<a name="line.894"></a> -<span class="sourceLineNo">895</span> LOG.warn("Could not find jar for class " + clazz +<a name="line.895"></a> -<span class="sourceLineNo">896</span> " in order to ship it to the cluster.");<a name="line.896"></a> -<span class="sourceLineNo">897</span> continue;<a name="line.897"></a> -<span class="sourceLineNo">898</span> }<a name="line.898"></a> -<span class="sourceLineNo">899</span> if (!localFs.exists(path)) {<a name="line.899"></a> -<span class="sourceLineNo">900</span> LOG.warn("Could not validate jar file " + path + " for class "<a name="line.900"></a> -<span class="sourceLineNo">901</span> + clazz);<a name="line.901"></a> -<span class="sourceLineNo">902</span> continue;<a name="line.902"></a> -<span class="sourceLineNo">903</span> }<a name="line.903"></a> -<span class="sourceLineNo">904</span> jars.add(path.toString());<a name="line.904"></a> -<span class="sourceLineNo">905</span> }<a name="line.905"></a> -<span class="sourceLineNo">906</span> if (jars.isEmpty()) return;<a name="line.906"></a> -<span class="sourceLineNo">907</span><a name="line.907"></a> -<span class="sourceLineNo">908</span> conf.set("tmpjars", StringUtils.arrayToString(jars.toArray(new String[jars.size()])));<a name="line.908"></a> -<span class="sourceLineNo">909</span> }<a name="line.909"></a> -<span class="sourceLineNo">910</span><a name="line.910"></a> -<span class="sourceLineNo">911</span> /**<a name="line.911"></a> -<span class="sourceLineNo">912</span> * Finds the Jar for a class or creates it if it doesn't exist. If the class is in<a name="line.912"></a> -<span class="sourceLineNo">913</span> * a directory in the classpath, it creates a Jar on the fly with the<a name="line.913"></a> -<span class="sourceLineNo">914</span> * contents of the directory and returns the path to that Jar. If a Jar is<a name="line.914"></a> -<span class="sourceLineNo">915</span> * created, it is created in the system temporary directory. Otherwise,<a name="line.915"></a> -<span class="sourceLineNo">916</span> * returns an existing jar that contains a class of the same name. Maintains<a name="line.916"></a> -<span class="sourceLineNo">917</span> * a mapping from jar contents to the tmp jar created.<a name="line.917"></a> -<span class="sourceLineNo">918</span> * @param my_class the class to find.<a name="line.918"></a> -<span class="sourceLineNo">919</span> * @param fs the FileSystem with which to qualify the returned path.<a name="line.919"></a> -<span class="sourceLineNo">920</span> * @param packagedClasses a map of class name to path.<a name="line.920"></a> -<span class="sourceLineNo">921</span> * @return a jar file that contains the class.<a name="line.921"></a> -<span class="sourceLineNo">922</span> * @throws IOException<a name="line.922"></a> -<span class="sourceLineNo">923</span> */<a name="line.923"></a> -<span class="sourceLineNo">924</span> private static Path findOrCreateJar(Class<?> my_class, FileSystem fs,<a name="line.924"></a> -<span class="sourceLineNo">925</span> Map<String, String> packagedClasses)<a name="line.925"></a> -<span class="sourceLineNo">926</span> throws IOException {<a name="line.926"></a> -<span class="sourceLineNo">927</span> // attempt to locate an existing jar for the class.<a name="line.927"></a> -<span class="sourceLineNo">928</span> String jar = findContainingJar(my_class, packagedClasses);<a name="line.928"></a> -<span class="sourceLineNo">929</span> if (null == jar || jar.isEmpty()) {<a name="line.929"></a> -<span class="sourceLineNo">930</span> jar = getJar(my_class);<a name="line.930"></a> -<span class="sourceLineNo">931</span> updateMap(jar, packagedClasses);<a name="line.931"></a> -<span class="sourceLineNo">932</span> }<a name="line.932"></a> -<span class="sourceLineNo">933</span><a name="line.933"></a> -<span class="sourceLineNo">934</span> if (null == jar || jar.isEmpty()) {<a name="line.934"></a> -<span class="sourceLineNo">935</span> return null;<a name="line.935"></a> -<span class="sourceLineNo">936</span> }<a name="line.936"></a> -<span class="sourceLineNo">937</span><a name="line.937"></a> -<span class="sourceLineNo">938</span> LOG.debug(String.format("For class %s, using jar %s", my_class.getName(), jar));<a name="line.938"></a> -<span class="sourceLineNo">939</span> return new Path(jar).makeQualified(fs);<a name="line.939"></a> -<span class="sourceLineNo">940</span> }<a name="line.940"></a> -<span class="sourceLineNo">941</span><a name="line.941"></a> -<span class="sourceLineNo">942</span> /**<a name="line.942"></a> -<span class="sourceLineNo">943</span> * Add entries to <code>packagedClasses</code> corresponding to class files<a name="line.943"></a> -<span class="sourceLineNo">944</span> * contained in <code>jar</code>.<a name="line.944"></a> -<span class="sourceLineNo">945</span> * @param jar The jar who's content to list.<a name="line.945"></a> -<span class="sourceLineNo">946</span> * @param packagedClasses map[class -> jar]<a name="line.946"></a> -<span class="sourceLineNo">947</span> */<a name="line.947"></a> -<span class="sourceLineNo">948</span> private static void updateMap(String jar, Map<String, String> packagedClasses) throws IOException {<a name="line.948"></a> -<span class="sourceLineNo">949</span> if (null == jar || jar.isEmpty()) {<a name="line.949"></a> -<span class="sourceLineNo">950</span> return;<a name="line.950"></a> -<span class="sourceLineNo">951</span> }<a name="line.951"></a> -<span class="sourceLineNo">952</span> ZipFile zip = null;<a name="line.952"></a> -<span class="sourceLineNo">953</span> try {<a name="line.953"></a> -<span class="sourceLineNo">954</span> zip = new ZipFile(jar);<a name="line.954"></a> -<span class="sourceLineNo">955</span> for (Enumeration<? extends ZipEntry> iter = zip.entries(); iter.hasMoreElements();) {<a name="line.955"></a> -<span class="sourceLineNo">956</span> ZipEntry entry = iter.nextElement();<a name="line.956"></a> -<span class="sourceLineNo">957</span> if (entry.getName().endsWith("class")) {<a name="line.957"></a> -<span class="sourceLineNo">958</span> packagedClasses.put(entry.getName(), jar);<a name="line.958"></a> -<span class="sourceLineNo">959</span> }<a name="line.959"></a> -<span class="sourceLineNo">960</span> }<a name="line.960"></a> -<span class="sourceLineNo">961</span> } finally {<a name="line.961"></a> -<span class="sourceLineNo">962</span> if (null != zip) zip.close();<a name="line.962"></a> -<span class="sourceLineNo">963</span> }<a name="line.963"></a> -<span class="sourceLineNo">964</span> }<a name="line.964"></a> -<span class="sourceLineNo">965</span><a name="line.965"></a> -<span class="sourceLineNo">966</span> /**<a name="line.966"></a> -<span class="sourceLineNo">967</span> * Find a jar that contains a class of the same name, if any. It will return<a name="line.967"></a> -<span class="sourceLineNo">968</span> * a jar file, even if that is not the first thing on the class path that<a name="line.968"></a> -<span class="sourceLineNo">969</span> * has a class with the same name. Looks first on the classpath and then in<a name="line.969"></a> -<span class="sourceLineNo">970</span> * the <code>packagedClasses</code> map.<a name="line.970"></a> -<span class="sourceLineNo">971</span> * @param my_class the class to find.<a name="line.971"></a> -<span class="sourceLineNo">972</span> * @return a jar file that contains the class, or null.<a name="line.972"></a> -<span class="sourceLineNo">973</span> * @throws IOException<a name="line.973"></a> -<span class="sourceLineNo">974</span> */<a name="line.974"></a> -<span class="sourceLineNo">975</span> private static String findContainingJar(Class<?> my_class, Map<String, String> packagedClasses)<a name="line.975"></a> -<span class="sourceLineNo">976</span> throws IOException {<a name="line.976"></a> -<span class="sourceLineNo">977</span> ClassLoader loader = my_class.getClassLoader();<a name="line.977"></a> -<span class="sourceLineNo">978</span><a name="line.978"></a> -<span class="sourceLineNo">979</span> String class_file = my_class.getName().replaceAll("\\.", "/") + ".class";<a name="line.979"></a> -<span class="sourceLineNo">980</span><a name="line.980"></a> -<span class="sourceLineNo">981</span> if (loader != null) {<a name="line.981"></a> -<span class="sourceLineNo">982</span> // first search the classpath<a name="line.982"></a> -<span class="sourceLineNo">983</span> for (Enumeration<URL> itr = loader.getResources(class_file); itr.hasMoreElements();) {<a name="line.983"></a> -<span class="sourceLineNo">984</span> URL url = itr.nextElement();<a name="line.984"></a> -<span class="sourceLineNo">985</span> if ("jar".equals(url.getProtocol())) {<a name="line.985"></a> -<span class="sourceLineNo">986</span> String toReturn = url.getPath();<a name="line.986"></a> -<span class="sourceLineNo">987</span> if (toReturn.startsWith("file:")) {<a name="line.987"></a> -<span class="sourceLineNo">988</span> toReturn = toReturn.substring("file:".length());<a name="line.988"></a> -<span class="sourceLineNo">989</span> }<a name="line.989"></a> -<span class="sourceLineNo">990</span> // URLDecoder is a misnamed class, since it actually decodes<a name="line.990"></a> -<span class="sourceLineNo">991</span> // x-www-form-urlencoded MIME type rather than actual<a name="line.991"></a> -<span class="sourceLineNo">992</span> // URL encoding (which the file path has). Therefore it would<a name="line.992"></a> -<span class="sourceLineNo">993</span> // decode +s to ' 's which is incorrect (spaces are actually<a name="line.993"></a> -<span class="sourceLineNo">994</span> // either unencoded or encoded as "%20"). Replace +s first, so<a name="line.994"></a> -<span class="sourceLineNo">995</span> // that they are kept sacred during the decoding process.<a name="line.995"></a> -<span class="sourceLineNo">996</span> toReturn = toReturn.replaceAll("\\+", "%2B");<a name="line.996"></a> -<span class="sourceLineNo">997</span> toReturn = URLDecoder.decode(toReturn, "UTF-8");<a name="line.997"></a> -<span class="sourceLineNo">998</span> return toReturn.replaceAll("!.*$", "");<a name="line.998"></a> -<span class="sourceLineNo">999</span> }<a name="line.999"></a> -<span class="sourceLineNo">1000</span> }<a name="line.1000"></a> -<span class="sourceLineNo">1001</span> }<a name="line.1001"></a> -<span class="sourceLineNo">1002</span><a name="line.1002"></a> -<span class="sourceLineNo">1003</span> // now look in any jars we've packaged using JarFinder. Returns null when<a name="line.1003"></a> -<span class="sourceLineNo">1004</span> // no jar is found.<a name="line.1004"></a> -<span class="sourceLineNo">1005</span> return packagedClasses.get(class_file);<a name="line.1005"></a> -<span class="sourceLineNo">1006</span> }<a name="line.1006"></a> -<span class="sourceLineNo">1007</span><a name="line.1007"></a> -<span class="sourceLineNo">1008</span> /**<a name="line.1008"></a> -<span class="sourceLineNo">1009</span> * Invoke 'getJar' on a custom JarFinder implementation. Useful for some job<a name="line.1009"></a> -<span class="sourceLineNo">1010</span> * configuration contexts (HBASE-8140) and also for testing on MRv2.<a name="line.1010"></a> -<span class="sourceLineNo">1011</span> * check if we have HADOOP-9426.<a name="line.1011"></a> -<span class="sourceLineNo">1012</span> * @param my_class the class to find.<a name="line.1012"></a> -<span class="sourceLineNo">1013</span> * @return a jar file that contains the class, or null.<a name="line.1013"></a> -<span class="sourceLineNo">1014</span> */<a name="line.1014"></a> -<span class="sourceLineNo">1015</span> private static String getJar(Class<?> my_class) {<a name="line.1015"></a> -<span class="sourceLineNo">1016</span> String ret = null;<a name="line.1016"></a> -<span class="sourceLineNo">1017</span> try {<a name="line.1017"></a> -<span class="sourceLineNo">1018</span> ret = JarFinder.getJar(my_class);<a name="line.1018"></a> -<span class="sourceLineNo">1019</span> } catch (Exception e) {<a name="line.1019"></a> -<span class="sourceLineNo">1020</span> // toss all other exceptions, related to reflection failure<a name="line.1020"></a> -<span class="sourceLineNo">1021</span> throw new RuntimeException("getJar invocation failed.", e);<a name="line.1021"></a> -<span class="sourceLineNo">1022</span> }<a name="line.1022"></a> -<span class="sourceLineNo">1023</span><a name="line.1023"></a> -<span class="sourceLineNo">1024</span> return ret;<a name="line.1024"></a> -<span class="sourceLineNo">1025</span> }<a name="line.1025"></a> -<span class="sourceLineNo">1026</span>}<a name="line.1026"></a> +<span class="sourceLineNo">791</span> org.apache.hadoop.hbase.metrics.impl.FastLongHistogram.class, // hbase-metrics<a name="line.791"></a> +<span class="sourceLineNo">792</span> prefixTreeCodecClass, // hbase-prefix-tree (if null will be skipped)<a name="line.792"></a> +<span class="sourceLineNo">793</span> // pull necessary dependencies<a name="line.793"></a> +<span class="sourceLineNo">794</span> org.apache.zookeeper.ZooKeeper.class,<a name="line.794"></a> +<span class="sourceLineNo">795</span> io.netty.channel.Channel.class,<a name="line.795"></a> +<span class="sourceLineNo">796</span> com.google.protobuf.Message.class,<a name="line.796"></a> +<span class="sourceLineNo">797</span> com.google.common.collect.Lists.class,<a name="line.797"></a> +<span class="sourceLineNo">798</span> org.apache.htrace.Trace.class,<a name="line.798"></a> +<span class="sourceLineNo">799</span> com.codahale.metrics.MetricRegistry.class);<a name="line.799"></a> +<span class="sourceLineNo">800</span> }<a name="line.800"></a> +<span class="sourceLineNo">801</span><a name="line.801"></a> +<span class="sourceLineNo">802</span> /**<a name="line.802"></a> +<span class="sourceLineNo">803</span> * Returns a classpath string built from the content of the "tmpjars" value in {@code conf}.<a name="line.803"></a> +<span class="sourceLineNo">804</span> * Also exposed to shell scripts via `bin/hbase mapredcp`.<a name="line.804"></a> +<span class="sourceLineNo">805</span> */<a name="line.805"></a> +<span class="sourceLineNo">806</span> public static String buildDependencyClasspath(Configuration conf) {<a name="line.806"></a> +<span class="sourceLineNo">807</span> if (conf == null) {<a name="line.807"></a> +<span class="sourceLineNo">808</span> throw new IllegalArgumentException("Must provide a configuration object.");<a name="line.808"></a> +<span class="sourceLineNo">809</span> }<a name="line.809"></a> +<span class="sourceLineNo">810</span> Set<String> paths = new HashSet<String>(conf.getStringCollection("tmpjars"));<a name="line.810"></a> +<span class="sourceLineNo">811</span> if (paths.isEmpty()) {<a name="line.811"></a> +<span class="sourceLineNo">812</span> throw new IllegalArgumentException("Configuration contains no tmpjars.");<a name="line.812"></a> +<span class="sourceLineNo">813</span> }<a name="line.813"></a> +<span class="sourceLineNo">814</span> StringBuilder sb = new StringBuilder();<a name="line.814"></a> +<span class="sourceLineNo">815</span> for (String s : paths) {<a name="line.815"></a> +<span class="sourceLineNo">816</span> // entries can take the form 'file:/path/to/file.jar'.<a name="line.816"></a> +<span class="sourceLineNo">817</span> int idx = s.indexOf(":");<a name="line.817"></a> +<span class="sourceLineNo">818</span> if (idx != -1) s = s.substring(idx + 1);<a name="line.818"></a> +<span class="sourceLineNo">819</span> if (sb.length() > 0) sb.append(File.pathSeparator);<a name="line.819"></a> +<span class="sourceLineNo">820</span> sb.append(s);<a name="line.820"></a> +<span class="sourceLineNo">821</span> }<a name="line.821"></a> +<span class="sourceLineNo">822</span> return sb.toString();<a name="line.822"></a> +<span class="sourceLineNo">823</span> }<a name="line.823"></a> +<span class="sourceLineNo">824</span><a name="line.824"></a> +<span class="sourceLineNo">825</span> /**<a name="line.825"></a> +<span class="sourceLineNo">826</span> * Add the HBase dependency jars as well as jars for any of the configured<a name="line.826"></a> +<span class="sourceLineNo">827</span> * job classes to the job configuration, so that JobClient will ship them<a name="line.827"></a> +<span class="sourceLineNo">828</span> * to the cluster and add them to the DistributedCache.<a name="line.828"></a> +<span class="sourceLineNo">829</span> */<a name="line.829"></a> +<span class="sourceLineNo">830</span> public static void addDependencyJars(Job job) throws IOException {<a name="line.830"></a> +<span class="sourceLineNo">831</span> addHBaseDependencyJars(job.getConfiguration());<a name="line.831"></a> +<span class="sourceLineNo">832</span> try {<a name="line.832"></a> +<span class="sourceLineNo">833</span> addDependencyJarsForClasses(job.getConfiguration(),<a name="line.833"></a> +<span class="sourceLineNo">834</span> // when making changes here, consider also mapred.TableMapReduceUtil<a name="line.834"></a> +<span class="sourceLineNo">835</span> // pull job classes<a name="line.835"></a> +<span class="sourceLineNo">836</span> job.getMapOutputKeyClass(),<a name="line.836"></a> +<span class="sourceLineNo">837</span> job.getMapOutputValueClass(),<a name="line.837"></a> +<span class="sourceLineNo">838</span> job.getInputFormatClass(),<a name="line.838"></a> +<span class="sourceLineNo">839</span> job.getOutputKeyClass(),<a name="line.839"></a> +<span class="sourceLineNo">840</span> job.getOutputValueClass(),<a name="line.840"></a> +<span class="sourceLineNo">841</span> job.getOutputFormatClass(),<a name="line.841"></a> +<span class="sourceLineNo">842</span> job.getPartitionerClass(),<a name="line.842"></a> +<span class="sourceLineNo">843</span> job.getCombinerClass());<a name="line.843"></a> +<span class="sourceLineNo">844</span> } catch (ClassNotFoundException e) {<a name="line.844"></a> +<span class="sourceLineNo">845</span> throw new IOException(e);<a name="line.845"></a> +<span class="sourceLineNo">846</span> }<a name="line.846"></a> +<span class="sourceLineNo">847</span> }<a name="line.847"></a> +<span class="sourceLineNo">848</span><a name="line.848"></a> +<span class="sourceLineNo">849</span> /**<a name="line.849"></a> +<span class="sourceLineNo">850</span> * Add the jars containing the given classes to the job's configuration<a name="line.850"></a> +<span class="sourceLineNo">851</span> * such that JobClient will ship them to the cluster and add them to<a name="line.851"></a> +<span class="sourceLineNo">852</span> * the DistributedCache.<a name="line.852"></a> +<span class="sourceLineNo">853</span> * @deprecated rely on {@link #addDependencyJars(Job)} instead.<a name="line.853"></a> +<span class="sourceLineNo">854</span> */<a name="line.854"></a> +<span class="sourceLineNo">855</span> @Deprecated<a name="line.855"></a> +<span class="sourceLineNo">856</span> public static void addDependencyJars(Configuration conf,<a name="line.856"></a> +<span class="sourceLineNo">857</span> Class<?>... classes) throws IOException {<a name="line.857"></a> +<span class="sourceLineNo">858</span> LOG.warn("The addDependencyJars(Configuration, Class<?>...) method has been deprecated since it"<a name="line.858"></a> +<span class="sourceLineNo">859</span> + " is easy to use incorrectly. Most users should rely on addDependencyJars(Job) " +<a name="line.859"></a> +<span class="sourceLineNo">860</span> "instead. See HBASE-8386 for more details.");<a name="line.860"></a> +<span class="sourceLineNo">861</span> addDependencyJarsForClasses(conf, classes);<a name="line.861"></a> +<span class="sourceLineNo">862</span> }<a name="line.862"></a> +<span class="sourceLineNo">863</span><a name="line.863"></a> +<span class="sourceLineNo">864</span> /**<a name="line.864"></a> +<span class="sourceLineNo">865</span> * Add the jars containing the given classes to the job's configuration<a name="line.865"></a> +<span class="sourceLineNo">866</span> * such that JobClient will ship them to the cluster and add them to<a name="line.866"></a> +<span class="sourceLineNo">867</span> * the DistributedCache.<a name="line.867"></a> +<span class="sourceLineNo">868</span> *<a name="line.868"></a> +<span class="sourceLineNo">869</span> * N.B. that this method at most adds one jar per class given. If there is more than one<a name="line.869"></a> +<span class="sourceLineNo">870</span> * jar available containing a class with the same name as a given class, we don't define<a name="line.870"></a> +<span class="sourceLineNo">871</span> * which of those jars might be chosen.<a name="line.871"></a> +<span class="sourceLineNo">872</span> *<a name="line.872"></a> +<span class="sourceLineNo">873</span> * @param conf The Hadoop Configuration to modify<a name="line.873"></a> +<span class="sourceLineNo">874</span> * @param classes will add just those dependencies needed to find the given classes<a name="line.874"></a> +<span class="sourceLineNo">875</span> * @throws IOException if an underlying library call fails.<a name="line.875"></a> +<span class="sourceLineNo">876</span> */<a name="line.876"></a> +<span class="sourceLineNo">877</span> @InterfaceAudience.Private<a name="line.877"></a> +<span class="sourceLineNo">878</span> public static void addDependencyJarsForClasses(Configuration conf,<a name="line.878"></a> +<span class="sourceLineNo">879</span> Class<?>... classes) throws IOException {<a name="line.879"></a> +<span class="sourceLineNo">880</span><a name="line.880"></a> +<span class="sourceLineNo">881</span> FileSystem localFs = FileSystem.getLocal(conf);<a name="line.881"></a> +<span class="sourceLineNo">882</span> Set<String> jars = new HashSet<String>();<a name="line.882"></a> +<span class="sourceLineNo">883</span> // Add jars that are already in the tmpjars variable<a name="line.883"></a> +<span class="sourceLineNo">884</span> jars.addAll(conf.getStringCollection("tmpjars"));<a name="line.884"></a> +<span class="sourceLineNo">885</span><a name="line.885"></a> +<span class="sourceLineNo">886</span> // add jars as we find them to a map of contents jar name so that we can avoid<a name="line.886"></a> +<span class="sourceLineNo">887</span> // creating new jars for classes that have already been packaged.<a name="line.887"></a> +<span class="sourceLineNo">888</span> Map<String, String> packagedClasses = new HashMap<String, String>();<a name="line.888"></a> +<span class="sourceLineNo">889</span><a name="line.889"></a> +<span class="sourceLineNo">890</span> // Add jars containing the specified classes<a name="line.890"></a> +<span class="sourceLineNo">891</span> for (Class<?> clazz : classes) {<a name="line.891"></a> +<span class="sourceLineNo">892</span> if (clazz == null) continue;<a name="line.892"></a> +<span class="sourceLineNo">893</span><a name="line.893"></a> +<span class="sourceLineNo">894</span> Path path = findOrCreateJar(clazz, localFs, packagedClasses);<a name="line.894"></a> +<span class="sourceLineNo">895</span> if (path == null) {<a name="line.895"></a> +<span class="sourceLineNo">896</span> LOG.warn("Could not find jar for class " + clazz +<a name="line.896"></a> +<span class="sourceLineNo">897</span> " in order to ship it to the cluster.");<a name="line.897"></a> +<span class="sourceLineNo">898</span> continue;<a name="line.898"></a> +<span class="sourceLineNo">899</span> }<a name="line.899"></a> +<span class="sourceLineNo">900</span> if (!localFs.exists(path)) {<a name="line.900"></a> +<span class="sourceLineNo">901</span> LOG.warn("Could not validate jar file " + path + " for class "<a name="line.901"></a> +<span class="sourceLineNo">902</span> + clazz);<a name="line.902"></a> +<span class="sourceLineNo">903</span> continue;<a name="line.903"></a> +<span class="sourceLineNo">904</span> }<a name="line.904"></a> +<span class="sourceLineNo">905</span> jars.add(path.toString());<a name="line.905"></a> +<span class="sourceLineNo">906</span> }<a name="line.906"></a> +<span class="sourceLineNo">907</span> if (jars.isEmpty()) return;<a name="line.907"></a> +<span class="sourceLineNo">908</span><a name="line.908"></a> +<span class="sourceLineNo">909</span> conf.set("tmpjars", StringUtils.arrayToString(jars.toArray(new String[jars.size()])));<a name="line.909"></a> +<span class="sourceLineNo">910</span> }<a name="line.910"></a> +<span class="sourceLineNo">911</span><a name="line.911"></a> +<span class="sourceLineNo">912</span> /**<a name="line.912"></a> +<span class="sourceLineNo">913</span> * Finds the Jar for a class or creates it if it doesn't exist. If the class is in<a name="line.913"></a> +<span class="sourceLineNo">914</span> * a directory in the classpath, it creates a Jar on the fly with the<a name="line.914"></a> +<span class="sourceLineNo">915</span> * contents of the directory and returns the path to that Jar. If a Jar is<a name="line.915"></a> +<span class="sourceLineNo">916</span> * created, it is created in the system temporary directory. Otherwise,<a name="line.916"></a> +<span class="sourceLineNo">917</span> * returns an existing jar that contains a class of the same name. Maintains<a name="line.917"></a> +<span class="sourceLineNo">918</span> * a mapping from jar contents to the tmp jar created.<a name="line.918"></a> +<span class="sourceLineNo">919</span> * @param my_class the class to find.<a name="line.919"></a> +<span class="sourceLineNo">920</span> * @param fs the FileSystem with which to qualify the returned path.<a name="line.920"></a> +<span class="sourceLineNo">921</span> * @param packagedClasses a map of class name to path.<a name="line.921"></a> +<span class="sourceLineNo">922</span> * @return a jar file that contains the class.<a name="line.922"></a> +<span class="sourceLineNo">923</span> * @throws IOException<a name="line.923"></a> +<span class="sourceLineNo">924</span> */<a name="line.924"></a> +<span class="sourceLineNo">925</span> private static Path findOrCreateJar(Class<?> my_class, FileSystem fs,<a name="line.925"></a> +<span class="sourceLineNo">926</span> Map<String, String> packagedClasses)<a name="line.926"></a> +<span class="sourceLineNo">927</span> throws IOException {<a name="line.927"></a> +<span class="sourceLineNo">928</span> // attempt to locate an existing jar for the class.<a name="line.928"></a> +<span class="sourceLineNo">929</span> String jar = findContainingJar(my_class, packagedClasses);<a name="line.929"></a> +<span class="sourceLineNo">930</span> if (null == jar || jar.isEmpty()) {<a name="line.930"></a> +<span class="sourceLineNo">931</span> jar = getJar(my_class);<a name="line.931"></a> +<span class="sourceLineNo">932</span> updateMap(jar, packagedClasses);<a name="line.932"></a> +<span class="sourceLineNo">933</span> }<a name="line.933"></a> +<span class="sourceLineNo">934</span><a name="line.934"></a> +<span class="sourceLineNo">935</span> if (null == jar || jar.isEmpty()) {<a name="line.935"></a> +<span class="sourceLineNo">936</span> return null;<a name="line.936"></a> +<span class="sourceLineNo">937</span> }<a name="line.937"></a> +<span class="sourceLineNo">938</span><a name="line.938"></a> +<span class="sourceLineNo">939</span> LOG.debug(String.format("For class %s, using jar %s", my_class.getName(), jar));<a name="line.939"></a> +<span class="sourceLineNo">940</span> return new Path(jar).makeQualified(fs);<a name="line.940"></a> +<span class="sourceLineNo">941</span> }<a name="line.941"></a> +<span class="sourceLineNo">942</span><a name="line.942"></a> +<span class="sourceLineNo">943</span> /**<a name="line.943"></a> +<span class="sourceLineNo">944</span> * Add entries to <code>packagedClasses</code> corresponding to class files<a name="line.944"></a> +<span class="sourceLineNo">945</span> * contained in <code>jar</code>.<a name="line.945"></a> +<span class="sourceLineNo">946</span> * @param jar The jar who's content to list.<a name="line.946"></a> +<span class="sourceLineNo">947</span> * @param packagedClasses map[class -> jar]<a name="line.947"></a> +<span class="sourceLineNo">948</span> */<a name="line.948"></a> +<span class="sourceLineNo">949</span> private static void updateMap(String jar, Map<String, String> packagedClasses) throws IOException {<a name="line.949"></a> +<span class="sourceLineNo">950</span> if (null == jar || jar.isEmpty()) {<a name="line.950"></a> +<span class="sourceLineNo">951</span> return;<a name="line.951"></a> +<span class="sourceLineNo">952</span> }<a name="line.952"></a> +<span class="sourceLineNo">953</span> ZipFile zip = null;<a name="line.953"></a> +<span class="sourceLineNo">954</span> try {<a name="line.954"></a> +<span class="sourceLineNo">955</span> zip = new ZipFile(jar);<a name="line.955"></a> +<span class="sourceLineNo">956</span> for (Enumeration<? extends ZipEntry> iter = zip.entries(); iter.hasMoreElements();) {<a name="line.956"></a> +<span class="sourceLineNo">957</span> ZipEntry entry = iter.nextElement();<a name="line.957"></a> +<span class="sourceLineNo">958</span> if (entry.getName().endsWith("class")) {<a name="line.958"></a> +<span class="sourceLineNo">959</span> packagedClasses.put(entry.getName(), jar);<a name="line.959"></a> +<span class="sourceLineNo">960</span> }<a name="line.960"></a> +<span class="sourceLineNo">961</span> }<a name="line.961"></a> +<span class="sourceLineNo">962</span> } finally {<a name="line.962"></a> +<span class="sourceLineNo">963</span> if (null != zip) zip.close();<a name="line.963"></a> +<span class="sourceLineNo">964</span> }<a name="line.964"></a> +<span class="sourceLineNo">965</span> }<a name="line.965"></a> +<span class="sourceLineNo">966</span><a name="line.966"></a> +<span class="sourceLineNo">967</span> /**<a name="line.967"></a> +<span class="sourceLineNo">968</span> * Find a jar that contains a class of the same name, if any. It will return<a name="line.968"></a> +<span class="sourceLineNo">969</span> * a jar file, even if that is not the first thing on the class path that<a name="line.969"></a> +<span class="sourceLineNo">970</span> * has a class with the same name. Looks first on the classpath and then in<a name="line.970"></a> +<span class="sourceLineNo">971</span> * the <code>packagedClasses</code> map.<a name="line.971"></a> +<span class="sourceLineNo">972</span> * @param my_class the class to find.<a name="line.972"></a> +<span class="sourceLineNo">973</span> * @return a jar file that contains the class, or null.<a name="line.973"></a> +<span class="sourceLineNo">974</span> * @throws IOException<a name="line.974"></a> +<span class="sourceLineNo">975</span> */<a name="line.975"></a> +<span class="sourceLineNo">976</span> private static String findContainingJar(Class<?> my_class, Map<String, String> packagedClasses)<a name="line.976"></a> +<span class="sourceLineNo">977</span> throws IOException {<a name="line.977"></a> +<span class="sourceLineNo">978</span> ClassLoader loader = my_class.getClassLoader();<a name="line.978"></a> +<span class="sourceLineNo">979</span><a name="line.979"></a> +<span class="sourceLineNo">980</span> String class_file = my_class.getName().replaceAll("\\.", "/") + ".class";<a name="line.980"></a> +<span class="sourceLineNo">981</span><a name="line.981"></a> +<span class="sourceLineNo">982</span> if (loader != null) {<a name="line.982"></a> +<span class="sourceLineNo">983</span> // first search the classpath<a name="line.983"></a> +<span class="sourceLineNo">984</span> for (Enumeration<URL> itr = loader.getResources(class_file); itr.hasMoreElements();) {<a name="line.984"></a> +<span class="sourceLineNo">985</span> URL url = itr.nextElement();<a name="line.985"></a> +<span class="sourceLineNo">986</span> if ("jar".equals(url.getProtocol())) {<a name="line.986"></a> +<span class="sourceLineNo">987</span> String toReturn = url.getPath();<a name="line.987"></a> +<span class="sourceLineNo">988</span> if (toReturn.startsWith("file:")) {<a name="line.988"></a> +<span class="sourceLineNo">989</span> toReturn = toReturn.substring("file:".length());<a name="line.989"></a> +<span class="sourceLineNo">990</span> }<a name="line.990"></a> +<span class="sourceLineNo">991</span> // URLDecoder is a misnamed class, since it actually decodes<a name="line.991"></a> +<span class="sourceLineNo">992</span> // x-www-form-urlencoded MIME type rather than actual<a name="line.992"></a> +<span class="sourceLineNo">993</span> // URL encoding (which the file path has). Therefore it would<a name="line.993"></a> +<span class="sourceLineNo">994</span> // decode +s to ' 's which is incorrect (spaces are actually<a name="line.994"></a> +<span class="sourceLineNo">995</span> // either unencoded or encoded as "%20"). Replace +s first, so<a name="line.995"></a> +<span class="sourceLineNo">996</span> // that they are kept sacred during the decoding process.<a name="line.996"></a> +<span class="sourceLineNo">997</span> toReturn = toReturn.replaceAll("\\+", "%2B");<a name="line.997"></a> +<span class="sourceLineNo">998</span> toReturn = URLDecoder.decode(toReturn, "UTF-8");<a name="line.998"></a> +<span class="sourceLineNo">999</span> return toReturn.replaceAll("!.*$", "");<a name="line.999"></a> +<span class="sourceLineNo">1000</span> }<a name="line.1000"></a> +<span class="sourceLineNo">1001</span> }<a name="line.1001"></a> +<span class="sourceLineNo">1002</span> }<a name="line.1002"></a> +<span class="sourceLineNo">1003</span><a name="line.1003"></a> +<span class="sourceLineNo">1004</span> // now look in any jars we've packaged using JarFinder. Returns null when<a name="line.1004"></a> +<span class="sourceLineNo">1005</span> // no jar is found.<a name="line.1005"></a> +<span class="sourceLineNo">1006</span> return packagedClasses.get(class_file);<a name="line.1006"></a> +<span class="sourceLineNo">1007</span> }<a name="line.1007"></a> +<span class="sourceLineNo">1008</span><a name="line.1008"></a> +<span class="sourceLineNo">1009</span> /**<a name="line.1009"></a> +<span class="sourceLineNo">1010</span> * Invoke 'getJar' on a custom JarFinder implementation. Useful for some job<a name="line.1010"></a> +<span class="sourceLineNo">1011</span> * configuration contexts (HBASE-8140) and also for testing on MRv2.<a name="line.1011"></a> +<span class="sourceLineNo">1012</span> * check if we have HADOOP-9426.<a name="line.1012"></a> +<span class="sourceLineNo">1013</span> * @param my_class the class to find.<a name="line.1013"></a> +<span class="sourceLineNo">1014</span> * @return a jar file that contains the class, or null.<a name="line.1014"></a> +<span class="sourceLineNo">1015</span> */<a name="line.1015"></a> +<span class="sourceLineNo">1016</span> private static String getJar(Class<?> my_class) {<a name="line.1016"></a> +<span class="sourceLineNo">1017</span> String ret = null;<a name="line.1017"></a> +<span class="sourceLineNo">1018</span> try {<a name="line.1018"></a> +<span class="sourceLineNo">1019</span> ret = JarFinder.getJar(my_class);<a name="line.1019"></a> +<span class="sourceLineNo">1020</span> } catch (Exception e) {<a name="line.1020"></a> +<span class="sourceLineNo">1021</span> // toss all other exceptions, related to reflection failure<a name="line.1021"></a> +<span class="sourceLineNo">1022</span> throw new RuntimeException("getJar invocation failed.", e);<a name="line.1022"></a> +<span class="sourceLineNo">1023</span> }<a name="line.1023"></a> +<span class="sourceLineNo">1024</span><a name="line.1024"></a> +<span class="sourceLineNo">1025</span> return ret;<a name="line.1025"></a> +<span class="sourceLineNo">1026</span> }<a name="line.1026"></a> +<span class="sourceLineNo">1027</span>}<a name="line.1027"></a>
