http://git-wip-us.apache.org/repos/asf/hbase-site/blob/9fb0764b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
----------------------------------------------------------------------
diff --git 
a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html 
b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
index e9af038..53cae9a 100644
--- 
a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
+++ 
b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
@@ -78,1244 +78,1245 @@
 <span class="sourceLineNo">070</span>import 
org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.70"></a>
 <span class="sourceLineNo">071</span>import 
org.apache.hadoop.hbase.client.Admin;<a name="line.71"></a>
 <span class="sourceLineNo">072</span>import 
org.apache.hadoop.hbase.client.ClientServiceCallable;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import 
org.apache.hadoop.hbase.client.Connection;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import 
org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import 
org.apache.hadoop.hbase.client.RegionLocator;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import 
org.apache.hadoop.hbase.client.RpcRetryingCallerFactory;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import 
org.apache.hadoop.hbase.client.SecureBulkLoadClient;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import 
org.apache.hadoop.hbase.client.Table;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import 
org.apache.hadoop.hbase.coprocessor.CoprocessorHost;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import 
org.apache.hadoop.hbase.io.HFileLink;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import 
org.apache.hadoop.hbase.io.HalfStoreFileReader;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import 
org.apache.hadoop.hbase.io.Reference;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import 
org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import 
org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import 
org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.85"></a>
-<span class="sourceLineNo">086</span>import 
org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.86"></a>
-<span class="sourceLineNo">087</span>import 
org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.87"></a>
-<span class="sourceLineNo">088</span>import 
org.apache.hadoop.hbase.io.hfile.HFileDataBlockEncoder;<a name="line.88"></a>
-<span class="sourceLineNo">089</span>import 
org.apache.hadoop.hbase.io.hfile.HFileScanner;<a name="line.89"></a>
-<span class="sourceLineNo">090</span>import 
org.apache.hadoop.hbase.ipc.RpcControllerFactory;<a name="line.90"></a>
-<span class="sourceLineNo">091</span>import 
org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.91"></a>
-<span class="sourceLineNo">092</span>import 
org.apache.hadoop.hbase.regionserver.HStore;<a name="line.92"></a>
-<span class="sourceLineNo">093</span>import 
org.apache.hadoop.hbase.regionserver.StoreFileInfo;<a name="line.93"></a>
-<span class="sourceLineNo">094</span>import 
org.apache.hadoop.hbase.regionserver.StoreFileWriter;<a name="line.94"></a>
-<span class="sourceLineNo">095</span>import 
org.apache.hadoop.hbase.security.UserProvider;<a name="line.95"></a>
-<span class="sourceLineNo">096</span>import 
org.apache.hadoop.hbase.security.token.FsDelegationToken;<a name="line.96"></a>
-<span class="sourceLineNo">097</span>import 
org.apache.hadoop.hbase.util.Bytes;<a name="line.97"></a>
-<span class="sourceLineNo">098</span>import 
org.apache.hadoop.hbase.util.FSHDFSUtils;<a name="line.98"></a>
-<span class="sourceLineNo">099</span>import 
org.apache.hadoop.hbase.util.Pair;<a name="line.99"></a>
-<span class="sourceLineNo">100</span>import org.apache.hadoop.util.Tool;<a 
name="line.100"></a>
-<span class="sourceLineNo">101</span>import 
org.apache.hadoop.util.ToolRunner;<a name="line.101"></a>
-<span class="sourceLineNo">102</span><a name="line.102"></a>
-<span class="sourceLineNo">103</span>/**<a name="line.103"></a>
-<span class="sourceLineNo">104</span> * Tool to load the output of 
HFileOutputFormat into an existing table.<a name="line.104"></a>
-<span class="sourceLineNo">105</span> */<a name="line.105"></a>
-<span class="sourceLineNo">106</span>@InterfaceAudience.Public<a 
name="line.106"></a>
-<span class="sourceLineNo">107</span>public class LoadIncrementalHFiles 
extends Configured implements Tool {<a name="line.107"></a>
-<span class="sourceLineNo">108</span>  private static final Log LOG = 
LogFactory.getLog(LoadIncrementalHFiles.class);<a name="line.108"></a>
-<span class="sourceLineNo">109</span>  private boolean initalized = false;<a 
name="line.109"></a>
-<span class="sourceLineNo">110</span><a name="line.110"></a>
-<span class="sourceLineNo">111</span>  public static final String NAME = 
"completebulkload";<a name="line.111"></a>
-<span class="sourceLineNo">112</span>  static final String 
RETRY_ON_IO_EXCEPTION = "hbase.bulkload.retries.retryOnIOException";<a 
name="line.112"></a>
-<span class="sourceLineNo">113</span>  public static final String 
MAX_FILES_PER_REGION_PER_FAMILY<a name="line.113"></a>
-<span class="sourceLineNo">114</span>    = 
"hbase.mapreduce.bulkload.max.hfiles.perRegion.perFamily";<a 
name="line.114"></a>
-<span class="sourceLineNo">115</span>  private static final String 
ASSIGN_SEQ_IDS = "hbase.mapreduce.bulkload.assign.sequenceNumbers";<a 
name="line.115"></a>
-<span class="sourceLineNo">116</span>  public final static String 
CREATE_TABLE_CONF_KEY = "create.table";<a name="line.116"></a>
-<span class="sourceLineNo">117</span>  public final static String 
IGNORE_UNMATCHED_CF_CONF_KEY = "ignore.unmatched.families";<a 
name="line.117"></a>
-<span class="sourceLineNo">118</span>  public final static String 
ALWAYS_COPY_FILES = "always.copy.files";<a name="line.118"></a>
-<span class="sourceLineNo">119</span><a name="line.119"></a>
-<span class="sourceLineNo">120</span>  // We use a '.' prefix which is ignored 
when walking directory trees<a name="line.120"></a>
-<span class="sourceLineNo">121</span>  // above. It is invalid family name.<a 
name="line.121"></a>
-<span class="sourceLineNo">122</span>  final static String TMP_DIR = ".tmp";<a 
name="line.122"></a>
-<span class="sourceLineNo">123</span><a name="line.123"></a>
-<span class="sourceLineNo">124</span>  private int 
maxFilesPerRegionPerFamily;<a name="line.124"></a>
-<span class="sourceLineNo">125</span>  private boolean assignSeqIds;<a 
name="line.125"></a>
-<span class="sourceLineNo">126</span>  private Set&lt;String&gt; 
unmatchedFamilies = new HashSet&lt;&gt;();<a name="line.126"></a>
-<span class="sourceLineNo">127</span><a name="line.127"></a>
-<span class="sourceLineNo">128</span>  // Source filesystem<a 
name="line.128"></a>
-<span class="sourceLineNo">129</span>  private FileSystem fs;<a 
name="line.129"></a>
-<span class="sourceLineNo">130</span>  // Source delegation token<a 
name="line.130"></a>
-<span class="sourceLineNo">131</span>  private FsDelegationToken 
fsDelegationToken;<a name="line.131"></a>
-<span class="sourceLineNo">132</span>  private String bulkToken;<a 
name="line.132"></a>
-<span class="sourceLineNo">133</span>  private UserProvider userProvider;<a 
name="line.133"></a>
-<span class="sourceLineNo">134</span>  private int nrThreads;<a 
name="line.134"></a>
-<span class="sourceLineNo">135</span>  private RpcControllerFactory 
rpcControllerFactory;<a name="line.135"></a>
-<span class="sourceLineNo">136</span>  private AtomicInteger numRetries;<a 
name="line.136"></a>
-<span class="sourceLineNo">137</span><a name="line.137"></a>
-<span class="sourceLineNo">138</span>  private Map&lt;LoadQueueItem, 
ByteBuffer&gt; retValue = null;<a name="line.138"></a>
-<span class="sourceLineNo">139</span><a name="line.139"></a>
-<span class="sourceLineNo">140</span>  public 
LoadIncrementalHFiles(Configuration conf) throws Exception {<a 
name="line.140"></a>
-<span class="sourceLineNo">141</span>    super(conf);<a name="line.141"></a>
-<span class="sourceLineNo">142</span>    this.rpcControllerFactory = new 
RpcControllerFactory(conf);<a name="line.142"></a>
-<span class="sourceLineNo">143</span>    initialize();<a name="line.143"></a>
-<span class="sourceLineNo">144</span>  }<a name="line.144"></a>
-<span class="sourceLineNo">145</span><a name="line.145"></a>
-<span class="sourceLineNo">146</span>  private void initialize() throws 
IOException {<a name="line.146"></a>
-<span class="sourceLineNo">147</span>    if (initalized) {<a 
name="line.147"></a>
-<span class="sourceLineNo">148</span>      return;<a name="line.148"></a>
-<span class="sourceLineNo">149</span>    }<a name="line.149"></a>
-<span class="sourceLineNo">150</span>    // make a copy, just to be sure we're 
not overriding someone else's config<a name="line.150"></a>
-<span class="sourceLineNo">151</span>    
setConf(HBaseConfiguration.create(getConf()));<a name="line.151"></a>
-<span class="sourceLineNo">152</span>    Configuration conf = getConf();<a 
name="line.152"></a>
-<span class="sourceLineNo">153</span>    // disable blockcache for tool 
invocation, see HBASE-10500<a name="line.153"></a>
-<span class="sourceLineNo">154</span>    
conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.154"></a>
-<span class="sourceLineNo">155</span>    this.userProvider = 
UserProvider.instantiate(conf);<a name="line.155"></a>
-<span class="sourceLineNo">156</span>    this.fsDelegationToken = new 
FsDelegationToken(userProvider, "renewer");<a name="line.156"></a>
-<span class="sourceLineNo">157</span>    assignSeqIds = 
conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.157"></a>
-<span class="sourceLineNo">158</span>    maxFilesPerRegionPerFamily = 
conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.158"></a>
-<span class="sourceLineNo">159</span>    nrThreads = 
conf.getInt("hbase.loadincremental.threads.max",<a name="line.159"></a>
-<span class="sourceLineNo">160</span>      
Runtime.getRuntime().availableProcessors());<a name="line.160"></a>
-<span class="sourceLineNo">161</span>    initalized = true;<a 
name="line.161"></a>
-<span class="sourceLineNo">162</span>    numRetries = new AtomicInteger(1);<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>  private void usage() {<a 
name="line.165"></a>
-<span class="sourceLineNo">166</span>    System.err.println("usage: " + NAME + 
" /path/to/hfileoutputformat-output tablename" + "\n -D"<a name="line.166"></a>
-<span class="sourceLineNo">167</span>        + CREATE_TABLE_CONF_KEY + "=no - 
can be used to avoid creation of table by this tool\n"<a name="line.167"></a>
-<span class="sourceLineNo">168</span>        + "  Note: if you set this to 
'no', then the target table must already exist in HBase\n -D"<a 
name="line.168"></a>
-<span class="sourceLineNo">169</span>        + IGNORE_UNMATCHED_CF_CONF_KEY + 
"=yes - can be used to ignore unmatched column families\n"<a 
name="line.169"></a>
-<span class="sourceLineNo">170</span>        + "\n");<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>  private interface 
BulkHFileVisitor&lt;TFamily&gt; {<a name="line.173"></a>
-<span class="sourceLineNo">174</span>    TFamily bulkFamily(final byte[] 
familyName)<a name="line.174"></a>
-<span class="sourceLineNo">175</span>      throws IOException;<a 
name="line.175"></a>
-<span class="sourceLineNo">176</span>    void bulkHFile(final TFamily family, 
final FileStatus hfileStatus)<a name="line.176"></a>
-<span class="sourceLineNo">177</span>      throws IOException;<a 
name="line.177"></a>
-<span class="sourceLineNo">178</span>  }<a name="line.178"></a>
-<span class="sourceLineNo">179</span><a name="line.179"></a>
-<span class="sourceLineNo">180</span>  /**<a name="line.180"></a>
-<span class="sourceLineNo">181</span>   * Iterate over the bulkDir hfiles.<a 
name="line.181"></a>
-<span class="sourceLineNo">182</span>   * Skip reference, HFileLink, files 
starting with "_" and non-valid hfiles.<a name="line.182"></a>
-<span class="sourceLineNo">183</span>   */<a name="line.183"></a>
-<span class="sourceLineNo">184</span>  private static &lt;TFamily&gt; void 
visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.184"></a>
-<span class="sourceLineNo">185</span>    final BulkHFileVisitor&lt;TFamily&gt; 
visitor) throws IOException {<a name="line.185"></a>
-<span class="sourceLineNo">186</span>    visitBulkHFiles(fs, bulkDir, visitor, 
true);<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>  /**<a name="line.189"></a>
-<span class="sourceLineNo">190</span>   * Iterate over the bulkDir hfiles.<a 
name="line.190"></a>
-<span class="sourceLineNo">191</span>   * Skip reference, HFileLink, files 
starting with "_".<a name="line.191"></a>
-<span class="sourceLineNo">192</span>   * Check and skip non-valid hfiles by 
default, or skip this validation by setting<a name="line.192"></a>
-<span class="sourceLineNo">193</span>   * 
'hbase.loadincremental.validate.hfile' to false.<a name="line.193"></a>
-<span class="sourceLineNo">194</span>   */<a name="line.194"></a>
-<span class="sourceLineNo">195</span>  private static &lt;TFamily&gt; void 
visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.195"></a>
-<span class="sourceLineNo">196</span>    final BulkHFileVisitor&lt;TFamily&gt; 
visitor, final boolean validateHFile) throws IOException {<a 
name="line.196"></a>
-<span class="sourceLineNo">197</span>    if (!fs.exists(bulkDir)) {<a 
name="line.197"></a>
-<span class="sourceLineNo">198</span>      throw new 
FileNotFoundException("Bulkload dir " + bulkDir + " not found");<a 
name="line.198"></a>
-<span class="sourceLineNo">199</span>    }<a name="line.199"></a>
-<span class="sourceLineNo">200</span><a name="line.200"></a>
-<span class="sourceLineNo">201</span>    FileStatus[] familyDirStatuses = 
fs.listStatus(bulkDir);<a name="line.201"></a>
-<span class="sourceLineNo">202</span>    if (familyDirStatuses == null) {<a 
name="line.202"></a>
-<span class="sourceLineNo">203</span>      throw new FileNotFoundException("No 
families found in " + bulkDir);<a name="line.203"></a>
-<span class="sourceLineNo">204</span>    }<a name="line.204"></a>
-<span class="sourceLineNo">205</span><a name="line.205"></a>
-<span class="sourceLineNo">206</span>    for (FileStatus familyStat : 
familyDirStatuses) {<a name="line.206"></a>
-<span class="sourceLineNo">207</span>      if (!familyStat.isDirectory()) {<a 
name="line.207"></a>
-<span class="sourceLineNo">208</span>        LOG.warn("Skipping non-directory 
" + familyStat.getPath());<a name="line.208"></a>
-<span class="sourceLineNo">209</span>        continue;<a name="line.209"></a>
-<span class="sourceLineNo">210</span>      }<a name="line.210"></a>
-<span class="sourceLineNo">211</span>      Path familyDir = 
familyStat.getPath();<a name="line.211"></a>
-<span class="sourceLineNo">212</span>      byte[] familyName = 
familyDir.getName().getBytes();<a name="line.212"></a>
-<span class="sourceLineNo">213</span>      // Skip invalid family<a 
name="line.213"></a>
-<span class="sourceLineNo">214</span>      try {<a name="line.214"></a>
-<span class="sourceLineNo">215</span>        
HColumnDescriptor.isLegalFamilyName(familyName);<a name="line.215"></a>
-<span class="sourceLineNo">216</span>      }<a name="line.216"></a>
-<span class="sourceLineNo">217</span>      catch (IllegalArgumentException e) 
{<a name="line.217"></a>
-<span class="sourceLineNo">218</span>        LOG.warn("Skipping invalid " + 
familyStat.getPath());<a name="line.218"></a>
-<span class="sourceLineNo">219</span>        continue;<a name="line.219"></a>
-<span class="sourceLineNo">220</span>      }<a name="line.220"></a>
-<span class="sourceLineNo">221</span>      TFamily family = 
visitor.bulkFamily(familyName);<a name="line.221"></a>
-<span class="sourceLineNo">222</span><a name="line.222"></a>
-<span class="sourceLineNo">223</span>      FileStatus[] hfileStatuses = 
fs.listStatus(familyDir);<a name="line.223"></a>
-<span class="sourceLineNo">224</span>      for (FileStatus hfileStatus : 
hfileStatuses) {<a name="line.224"></a>
-<span class="sourceLineNo">225</span>        if 
(!fs.isFile(hfileStatus.getPath())) {<a name="line.225"></a>
-<span class="sourceLineNo">226</span>          LOG.warn("Skipping non-file " + 
hfileStatus);<a name="line.226"></a>
-<span class="sourceLineNo">227</span>          continue;<a name="line.227"></a>
-<span class="sourceLineNo">228</span>        }<a name="line.228"></a>
-<span class="sourceLineNo">229</span><a name="line.229"></a>
-<span class="sourceLineNo">230</span>        Path hfile = 
hfileStatus.getPath();<a name="line.230"></a>
-<span class="sourceLineNo">231</span>        // Skip "_", reference, 
HFileLink<a name="line.231"></a>
-<span class="sourceLineNo">232</span>        String fileName = 
hfile.getName();<a name="line.232"></a>
-<span class="sourceLineNo">233</span>        if (fileName.startsWith("_")) {<a 
name="line.233"></a>
-<span class="sourceLineNo">234</span>          continue;<a name="line.234"></a>
-<span class="sourceLineNo">235</span>        }<a name="line.235"></a>
-<span class="sourceLineNo">236</span>        if 
(StoreFileInfo.isReference(fileName)) {<a name="line.236"></a>
-<span class="sourceLineNo">237</span>          LOG.warn("Skipping reference " 
+ fileName);<a name="line.237"></a>
-<span class="sourceLineNo">238</span>          continue;<a name="line.238"></a>
-<span class="sourceLineNo">239</span>        }<a name="line.239"></a>
-<span class="sourceLineNo">240</span>        if 
(HFileLink.isHFileLink(fileName)) {<a name="line.240"></a>
-<span class="sourceLineNo">241</span>          LOG.warn("Skipping HFileLink " 
+ fileName);<a name="line.241"></a>
-<span class="sourceLineNo">242</span>          continue;<a name="line.242"></a>
-<span class="sourceLineNo">243</span>        }<a name="line.243"></a>
-<span class="sourceLineNo">244</span><a name="line.244"></a>
-<span class="sourceLineNo">245</span>        // Validate HFile Format if 
needed<a name="line.245"></a>
-<span class="sourceLineNo">246</span>        if (validateHFile) {<a 
name="line.246"></a>
-<span class="sourceLineNo">247</span>          try {<a name="line.247"></a>
-<span class="sourceLineNo">248</span>            if (!HFile.isHFileFormat(fs, 
hfile)) {<a name="line.248"></a>
-<span class="sourceLineNo">249</span>              LOG.warn("the file " + 
hfile + " doesn't seems to be an hfile. skipping");<a name="line.249"></a>
-<span class="sourceLineNo">250</span>              continue;<a 
name="line.250"></a>
-<span class="sourceLineNo">251</span>            }<a name="line.251"></a>
-<span class="sourceLineNo">252</span>          } catch (FileNotFoundException 
e) {<a name="line.252"></a>
-<span class="sourceLineNo">253</span>            LOG.warn("the file " + hfile 
+ " was removed");<a name="line.253"></a>
-<span class="sourceLineNo">254</span>            continue;<a 
name="line.254"></a>
-<span class="sourceLineNo">255</span>          }<a name="line.255"></a>
-<span class="sourceLineNo">256</span>  }<a name="line.256"></a>
-<span class="sourceLineNo">257</span><a name="line.257"></a>
-<span class="sourceLineNo">258</span>        visitor.bulkHFile(family, 
hfileStatus);<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>
-<span class="sourceLineNo">262</span><a name="line.262"></a>
-<span class="sourceLineNo">263</span>  /**<a name="line.263"></a>
-<span class="sourceLineNo">264</span>   * Represents an HFile waiting to be 
loaded. An queue is used<a name="line.264"></a>
-<span class="sourceLineNo">265</span>   * in this class in order to support 
the case where a region has<a name="line.265"></a>
-<span class="sourceLineNo">266</span>   * split during the process of the 
load. When this happens,<a name="line.266"></a>
-<span class="sourceLineNo">267</span>   * the HFile is split into two physical 
parts across the new<a name="line.267"></a>
-<span class="sourceLineNo">268</span>   * region boundary, and each part is 
added back into the queue.<a name="line.268"></a>
-<span class="sourceLineNo">269</span>   * The import process finishes when the 
queue is empty.<a name="line.269"></a>
-<span class="sourceLineNo">270</span>   */<a name="line.270"></a>
-<span class="sourceLineNo">271</span>  public static class LoadQueueItem {<a 
name="line.271"></a>
-<span class="sourceLineNo">272</span>    final byte[] family;<a 
name="line.272"></a>
-<span class="sourceLineNo">273</span>    final Path hfilePath;<a 
name="line.273"></a>
-<span class="sourceLineNo">274</span><a name="line.274"></a>
-<span class="sourceLineNo">275</span>    public LoadQueueItem(byte[] family, 
Path hfilePath) {<a name="line.275"></a>
-<span class="sourceLineNo">276</span>      this.family = family;<a 
name="line.276"></a>
-<span class="sourceLineNo">277</span>      this.hfilePath = hfilePath;<a 
name="line.277"></a>
-<span class="sourceLineNo">278</span>    }<a name="line.278"></a>
-<span class="sourceLineNo">279</span><a name="line.279"></a>
-<span class="sourceLineNo">280</span>    @Override<a name="line.280"></a>
-<span class="sourceLineNo">281</span>    public String toString() {<a 
name="line.281"></a>
-<span class="sourceLineNo">282</span>      return "family:"+ 
Bytes.toString(family) + " path:" + hfilePath.toString();<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>    public byte[] getFamily() {<a 
name="line.285"></a>
-<span class="sourceLineNo">286</span>      return family;<a 
name="line.286"></a>
-<span class="sourceLineNo">287</span>    }<a name="line.287"></a>
-<span class="sourceLineNo">288</span><a name="line.288"></a>
-<span class="sourceLineNo">289</span>    public Path getFilePath() {<a 
name="line.289"></a>
-<span class="sourceLineNo">290</span>      return hfilePath;<a 
name="line.290"></a>
-<span class="sourceLineNo">291</span>    }<a name="line.291"></a>
-<span class="sourceLineNo">292</span>  }<a name="line.292"></a>
-<span class="sourceLineNo">293</span><a name="line.293"></a>
-<span class="sourceLineNo">294</span>  /*<a name="line.294"></a>
-<span class="sourceLineNo">295</span>   * Populate the Queue with given 
HFiles<a name="line.295"></a>
-<span class="sourceLineNo">296</span>   */<a name="line.296"></a>
-<span class="sourceLineNo">297</span>  private void populateLoadQueue(final 
Deque&lt;LoadQueueItem&gt; ret,<a name="line.297"></a>
-<span class="sourceLineNo">298</span>      Map&lt;byte[], List&lt;Path&gt;&gt; 
map) throws IOException {<a name="line.298"></a>
-<span class="sourceLineNo">299</span>    for (Map.Entry&lt;byte[], 
List&lt;Path&gt;&gt; entry : map.entrySet()) {<a name="line.299"></a>
-<span class="sourceLineNo">300</span>      for (Path p : entry.getValue()) {<a 
name="line.300"></a>
-<span class="sourceLineNo">301</span>        ret.add(new 
LoadQueueItem(entry.getKey(), p));<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>  }<a name="line.304"></a>
-<span class="sourceLineNo">305</span><a name="line.305"></a>
-<span class="sourceLineNo">306</span>  /**<a name="line.306"></a>
-<span class="sourceLineNo">307</span>   * Walk the given directory for all 
HFiles, and return a Queue<a name="line.307"></a>
-<span class="sourceLineNo">308</span>   * containing all such files.<a 
name="line.308"></a>
-<span class="sourceLineNo">309</span>   */<a name="line.309"></a>
-<span class="sourceLineNo">310</span>  private void discoverLoadQueue(final 
Deque&lt;LoadQueueItem&gt; ret, final Path hfofDir,<a name="line.310"></a>
-<span class="sourceLineNo">311</span>      final boolean validateHFile) throws 
IOException {<a name="line.311"></a>
-<span class="sourceLineNo">312</span>    fs = 
hfofDir.getFileSystem(getConf());<a name="line.312"></a>
-<span class="sourceLineNo">313</span>    visitBulkHFiles(fs, hfofDir, new 
BulkHFileVisitor&lt;byte[]&gt;() {<a name="line.313"></a>
-<span class="sourceLineNo">314</span>      @Override<a name="line.314"></a>
-<span class="sourceLineNo">315</span>      public byte[] bulkFamily(final 
byte[] familyName) {<a name="line.315"></a>
-<span class="sourceLineNo">316</span>        return familyName;<a 
name="line.316"></a>
-<span class="sourceLineNo">317</span>      }<a name="line.317"></a>
-<span class="sourceLineNo">318</span>      @Override<a name="line.318"></a>
-<span class="sourceLineNo">319</span>      public void bulkHFile(final byte[] 
family, final FileStatus hfile) throws IOException {<a name="line.319"></a>
-<span class="sourceLineNo">320</span>        long length = hfile.getLen();<a 
name="line.320"></a>
-<span class="sourceLineNo">321</span>        if (length &gt; 
getConf().getLong(HConstants.HREGION_MAX_FILESIZE,<a name="line.321"></a>
-<span class="sourceLineNo">322</span>            
HConstants.DEFAULT_MAX_FILE_SIZE)) {<a name="line.322"></a>
-<span class="sourceLineNo">323</span>          LOG.warn("Trying to bulk load 
hfile " + hfile.getPath() + " with size: " +<a name="line.323"></a>
-<span class="sourceLineNo">324</span>              length + " bytes can be 
problematic as it may lead to oversplitting.");<a name="line.324"></a>
-<span class="sourceLineNo">325</span>        }<a name="line.325"></a>
-<span class="sourceLineNo">326</span>        ret.add(new LoadQueueItem(family, 
hfile.getPath()));<a name="line.326"></a>
-<span class="sourceLineNo">327</span>      }<a name="line.327"></a>
-<span class="sourceLineNo">328</span>    }, validateHFile);<a 
name="line.328"></a>
-<span class="sourceLineNo">329</span>  }<a name="line.329"></a>
-<span class="sourceLineNo">330</span><a name="line.330"></a>
-<span class="sourceLineNo">331</span>  /**<a name="line.331"></a>
-<span class="sourceLineNo">332</span>   * Perform a bulk load of the given 
directory into the given<a name="line.332"></a>
-<span class="sourceLineNo">333</span>   * pre-existing table.  This method is 
not threadsafe.<a name="line.333"></a>
-<span class="sourceLineNo">334</span>   *<a name="line.334"></a>
-<span class="sourceLineNo">335</span>   * @param hfofDir the directory that 
was provided as the output path<a name="line.335"></a>
-<span class="sourceLineNo">336</span>   *   of a job using HFileOutputFormat<a 
name="line.336"></a>
-<span class="sourceLineNo">337</span>   * @param admin the Admin<a 
name="line.337"></a>
-<span class="sourceLineNo">338</span>   * @param table the table to load 
into<a name="line.338"></a>
-<span class="sourceLineNo">339</span>   * @param regionLocator region 
locator<a name="line.339"></a>
-<span class="sourceLineNo">340</span>   * @throws TableNotFoundException if 
table does not yet exist<a name="line.340"></a>
-<span class="sourceLineNo">341</span>   */<a name="line.341"></a>
-<span class="sourceLineNo">342</span>  public void doBulkLoad(Path hfofDir, 
final Admin admin, Table table,<a name="line.342"></a>
-<span class="sourceLineNo">343</span>      RegionLocator regionLocator) throws 
TableNotFoundException, IOException {<a name="line.343"></a>
-<span class="sourceLineNo">344</span>    doBulkLoad(hfofDir, admin, table, 
regionLocator, false, false);<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>  void cleanup(Admin admin, 
Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.347"></a>
-<span class="sourceLineNo">348</span>      SecureBulkLoadClient secureClient) 
throws IOException {<a name="line.348"></a>
-<span class="sourceLineNo">349</span>    
fsDelegationToken.releaseDelegationToken();<a name="line.349"></a>
-<span class="sourceLineNo">350</span>    if (bulkToken != null &amp;&amp; 
secureClient != null) {<a name="line.350"></a>
-<span class="sourceLineNo">351</span>      
secureClient.cleanupBulkLoad(admin.getConnection(), bulkToken);<a 
name="line.351"></a>
-<span class="sourceLineNo">352</span>    }<a name="line.352"></a>
-<span class="sourceLineNo">353</span>    if (pool != null) {<a 
name="line.353"></a>
-<span class="sourceLineNo">354</span>      pool.shutdown();<a 
name="line.354"></a>
-<span class="sourceLineNo">355</span>    }<a name="line.355"></a>
-<span class="sourceLineNo">356</span>    if (!queue.isEmpty()) {<a 
name="line.356"></a>
-<span class="sourceLineNo">357</span>      StringBuilder err = new 
StringBuilder();<a name="line.357"></a>
-<span class="sourceLineNo">358</span>      
err.append("-------------------------------------------------\n");<a 
name="line.358"></a>
-<span class="sourceLineNo">359</span>      err.append("Bulk load aborted with 
some files not yet loaded:\n");<a name="line.359"></a>
-<span class="sourceLineNo">360</span>      
err.append("-------------------------------------------------\n");<a 
name="line.360"></a>
-<span class="sourceLineNo">361</span>      for (LoadQueueItem q : queue) {<a 
name="line.361"></a>
-<span class="sourceLineNo">362</span>        err.append("  
").append(q.hfilePath).append('\n');<a name="line.362"></a>
-<span class="sourceLineNo">363</span>      }<a name="line.363"></a>
-<span class="sourceLineNo">364</span>      LOG.error(err);<a 
name="line.364"></a>
-<span class="sourceLineNo">365</span>    }<a name="line.365"></a>
-<span class="sourceLineNo">366</span>  }<a name="line.366"></a>
-<span class="sourceLineNo">367</span>  /**<a name="line.367"></a>
-<span class="sourceLineNo">368</span>   * Perform a bulk load of the given 
directory into the given<a name="line.368"></a>
-<span class="sourceLineNo">369</span>   * pre-existing table.  This method is 
not threadsafe.<a name="line.369"></a>
-<span class="sourceLineNo">370</span>   *<a name="line.370"></a>
-<span class="sourceLineNo">371</span>   * @param map map of family to List of 
hfiles<a name="line.371"></a>
-<span class="sourceLineNo">372</span>   * @param admin the Admin<a 
name="line.372"></a>
-<span class="sourceLineNo">373</span>   * @param table the table to load 
into<a name="line.373"></a>
-<span class="sourceLineNo">374</span>   * @param regionLocator region 
locator<a name="line.374"></a>
-<span class="sourceLineNo">375</span>   * @param silence true to ignore 
unmatched column families<a name="line.375"></a>
-<span class="sourceLineNo">376</span>   * @param copyFile always copy hfiles 
if true<a name="line.376"></a>
-<span class="sourceLineNo">377</span>   * @throws TableNotFoundException if 
table does not yet exist<a name="line.377"></a>
-<span class="sourceLineNo">378</span>   */<a name="line.378"></a>
-<span class="sourceLineNo">379</span>  public void doBulkLoad(Map&lt;byte[], 
List&lt;Path&gt;&gt; map, final Admin admin,<a name="line.379"></a>
-<span class="sourceLineNo">380</span>      Table table, RegionLocator 
regionLocator, boolean silence, boolean copyFile)<a name="line.380"></a>
-<span class="sourceLineNo">381</span>          throws TableNotFoundException, 
IOException {<a name="line.381"></a>
-<span class="sourceLineNo">382</span>    if 
(!admin.isTableAvailable(regionLocator.getName())) {<a name="line.382"></a>
-<span class="sourceLineNo">383</span>      throw new 
TableNotFoundException("Table " + table.getName() + " is not currently 
available.");<a name="line.383"></a>
-<span class="sourceLineNo">384</span>    }<a name="line.384"></a>
-<span class="sourceLineNo">385</span>    // LQI queue does not need to be 
threadsafe -- all operations on this queue<a name="line.385"></a>
-<span class="sourceLineNo">386</span>    // happen in this thread<a 
name="line.386"></a>
-<span class="sourceLineNo">387</span>    Deque&lt;LoadQueueItem&gt; queue = 
new LinkedList&lt;&gt;();<a name="line.387"></a>
-<span class="sourceLineNo">388</span>    ExecutorService pool = null;<a 
name="line.388"></a>
-<span class="sourceLineNo">389</span>    SecureBulkLoadClient secureClient = 
null;<a name="line.389"></a>
-<span class="sourceLineNo">390</span>    try {<a name="line.390"></a>
-<span class="sourceLineNo">391</span>      prepareHFileQueue(map, table, 
queue, silence);<a name="line.391"></a>
-<span class="sourceLineNo">392</span>      if (queue.isEmpty()) {<a 
name="line.392"></a>
-<span class="sourceLineNo">393</span>        LOG.warn("Bulk load operation did 
not get any files to load");<a name="line.393"></a>
-<span class="sourceLineNo">394</span>        return;<a name="line.394"></a>
-<span class="sourceLineNo">395</span>      }<a name="line.395"></a>
-<span class="sourceLineNo">396</span>      pool = createExecutorService();<a 
name="line.396"></a>
-<span class="sourceLineNo">397</span>      secureClient = new 
SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.397"></a>
-<span class="sourceLineNo">398</span>      for (Map.Entry&lt;byte[], 
List&lt;Path&gt;&gt; entry : map.entrySet()) {<a name="line.398"></a>
-<span class="sourceLineNo">399</span>        for (Path p : entry.getValue()) 
{<a name="line.399"></a>
-<span class="sourceLineNo">400</span>          fs = 
p.getFileSystem(table.getConfiguration());<a name="line.400"></a>
-<span class="sourceLineNo">401</span>          break;<a name="line.401"></a>
-<span class="sourceLineNo">402</span>        }<a name="line.402"></a>
-<span class="sourceLineNo">403</span>      }<a name="line.403"></a>
-<span class="sourceLineNo">404</span>      retValue = performBulkLoad(admin, 
table, regionLocator, queue, pool, secureClient, copyFile);<a 
name="line.404"></a>
-<span class="sourceLineNo">405</span>    } finally {<a name="line.405"></a>
-<span class="sourceLineNo">406</span>      cleanup(admin, queue, pool, 
secureClient);<a name="line.406"></a>
-<span class="sourceLineNo">407</span>    }<a name="line.407"></a>
-<span class="sourceLineNo">408</span>  }<a name="line.408"></a>
-<span class="sourceLineNo">409</span><a name="line.409"></a>
-<span class="sourceLineNo">410</span>  /**<a name="line.410"></a>
-<span class="sourceLineNo">411</span>   * Perform a bulk load of the given 
directory into the given<a name="line.411"></a>
-<span class="sourceLineNo">412</span>   * pre-existing table.  This method is 
not threadsafe.<a name="line.412"></a>
-<span class="sourceLineNo">413</span>   *<a name="line.413"></a>
-<span class="sourceLineNo">414</span>   * @param hfofDir the directory that 
was provided as the output path<a name="line.414"></a>
-<span class="sourceLineNo">415</span>   *   of a job using HFileOutputFormat<a 
name="line.415"></a>
-<span class="sourceLineNo">416</span>   * @param admin the Admin<a 
name="line.416"></a>
-<span class="sourceLineNo">417</span>   * @param table the table to load 
into<a name="line.417"></a>
-<span class="sourceLineNo">418</span>   * @param regionLocator region 
locator<a name="line.418"></a>
-<span class="sourceLineNo">419</span>   * @param silence true to ignore 
unmatched column families<a name="line.419"></a>
-<span class="sourceLineNo">420</span>   * @param copyFile always copy hfiles 
if true<a name="line.420"></a>
-<span class="sourceLineNo">421</span>   * @throws TableNotFoundException if 
table does not yet exist<a name="line.421"></a>
-<span class="sourceLineNo">422</span>   */<a name="line.422"></a>
-<span class="sourceLineNo">423</span>  public void doBulkLoad(Path hfofDir, 
final Admin admin, Table table,<a name="line.423"></a>
-<span class="sourceLineNo">424</span>      RegionLocator regionLocator, 
boolean silence, boolean copyFile)<a name="line.424"></a>
-<span class="sourceLineNo">425</span>          throws TableNotFoundException, 
IOException {<a name="line.425"></a>
-<span class="sourceLineNo">426</span>    if 
(!admin.isTableAvailable(regionLocator.getName())) {<a name="line.426"></a>
-<span class="sourceLineNo">427</span>      throw new 
TableNotFoundException("Table " + table.getName() + " is not currently 
available.");<a name="line.427"></a>
-<span class="sourceLineNo">428</span>    }<a name="line.428"></a>
-<span class="sourceLineNo">429</span><a name="line.429"></a>
-<span class="sourceLineNo">430</span>    /*<a name="line.430"></a>
-<span class="sourceLineNo">431</span>     * Checking hfile format is a 
time-consuming operation, we should have an option to skip<a 
name="line.431"></a>
-<span class="sourceLineNo">432</span>     * this step when bulkloading 
millions of HFiles. See HBASE-13985.<a name="line.432"></a>
-<span class="sourceLineNo">433</span>     */<a name="line.433"></a>
-<span class="sourceLineNo">434</span>    boolean validateHFile = 
getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a 
name="line.434"></a>
-<span class="sourceLineNo">435</span>    if (!validateHFile) {<a 
name="line.435"></a>
-<span class="sourceLineNo">436</span>      LOG.warn("You are skipping HFiles 
validation, it might cause some data loss if files " +<a name="line.436"></a>
-<span class="sourceLineNo">437</span>          "are not correct. If you fail 
to read data from your table after using this " +<a name="line.437"></a>
-<span class="sourceLineNo">438</span>          "option, consider removing the 
files and bulkload again without this option. " +<a name="line.438"></a>
-<span class="sourceLineNo">439</span>          "See HBASE-13985");<a 
name="line.439"></a>
-<span class="sourceLineNo">440</span>    }<a name="line.440"></a>
-<span class="sourceLineNo">441</span>    // LQI queue does not need to be 
threadsafe -- all operations on this queue<a name="line.441"></a>
-<span class="sourceLineNo">442</span>    // happen in this thread<a 
name="line.442"></a>
-<span class="sourceLineNo">443</span>    Deque&lt;LoadQueueItem&gt; queue = 
new LinkedList&lt;&gt;();<a name="line.443"></a>
-<span class="sourceLineNo">444</span>    ExecutorService pool = null;<a 
name="line.444"></a>
-<span class="sourceLineNo">445</span>    SecureBulkLoadClient secureClient = 
null;<a name="line.445"></a>
-<span class="sourceLineNo">446</span>    try {<a name="line.446"></a>
-<span class="sourceLineNo">447</span>      prepareHFileQueue(hfofDir, table, 
queue, validateHFile, silence);<a name="line.447"></a>
-<span class="sourceLineNo">448</span><a name="line.448"></a>
-<span class="sourceLineNo">449</span>      if (queue.isEmpty()) {<a 
name="line.449"></a>
-<span class="sourceLineNo">450</span>        LOG.warn("Bulk load operation did 
not find any files to load in " +<a name="line.450"></a>
-<span class="sourceLineNo">451</span>            "directory " + hfofDir != 
null ? hfofDir.toUri() : "" + ".  Does it contain files in " +<a 
name="line.451"></a>
-<span class="sourceLineNo">452</span>            "subdirectories that 
correspond to column family names?");<a name="line.452"></a>
-<span class="sourceLineNo">453</span>        return;<a name="line.453"></a>
-<span class="sourceLineNo">454</span>      }<a name="line.454"></a>
-<span class="sourceLineNo">455</span>      pool = createExecutorService();<a 
name="line.455"></a>
-<span class="sourceLineNo">456</span>      secureClient = new 
SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.456"></a>
-<span class="sourceLineNo">457</span>      retValue = performBulkLoad(admin, 
table, regionLocator, queue, pool, secureClient, copyFile);<a 
name="line.457"></a>
-<span class="sourceLineNo">458</span>    } finally {<a name="line.458"></a>
-<span class="sourceLineNo">459</span>      cleanup(admin, queue, pool, 
secureClient);<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>  Map&lt;LoadQueueItem, ByteBuffer&gt; 
performBulkLoad(final Admin admin, Table table,<a name="line.463"></a>
-<span class="sourceLineNo">464</span>      RegionLocator regionLocator, 
Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.464"></a>
-<span class="sourceLineNo">465</span>      SecureBulkLoadClient secureClient, 
boolean copyFile) throws IOException {<a name="line.465"></a>
-<span class="sourceLineNo">466</span>    int count = 0;<a name="line.466"></a>
-<span class="sourceLineNo">467</span><a name="line.467"></a>
-<span class="sourceLineNo">468</span>    
if(isSecureBulkLoadEndpointAvailable()) {<a name="line.468"></a>
-<span class="sourceLineNo">469</span>      LOG.warn("SecureBulkLoadEndpoint is 
deprecated. It will be removed in future releases.");<a name="line.469"></a>
-<span class="sourceLineNo">470</span>      LOG.warn("Secure bulk load has been 
integrated into HBase core.");<a name="line.470"></a>
-<span class="sourceLineNo">471</span>    }<a name="line.471"></a>
-<span class="sourceLineNo">472</span><a name="line.472"></a>
-<span class="sourceLineNo">473</span>    //If using secure bulk load, get 
source delegation token, and<a name="line.473"></a>
-<span class="sourceLineNo">474</span>    //prepare staging directory and 
token<a name="line.474"></a>
-<span class="sourceLineNo">475</span>    // fs is the source filesystem<a 
name="line.475"></a>
-<span class="sourceLineNo">476</span>    
fsDelegationToken.acquireDelegationToken(fs);<a name="line.476"></a>
-<span class="sourceLineNo">477</span>    bulkToken = 
secureClient.prepareBulkLoad(admin.getConnection());<a name="line.477"></a>
-<span class="sourceLineNo">478</span>    Pair&lt;Multimap&lt;ByteBuffer, 
LoadQueueItem&gt;, Set&lt;String&gt;&gt; pair = null;<a name="line.478"></a>
-<span class="sourceLineNo">479</span><a name="line.479"></a>
-<span class="sourceLineNo">480</span>    Map&lt;LoadQueueItem, ByteBuffer&gt; 
item2RegionMap = new HashMap&lt;&gt;();<a name="line.480"></a>
-<span class="sourceLineNo">481</span>    // Assumes that region splits can 
happen while this occurs.<a name="line.481"></a>
-<span class="sourceLineNo">482</span>    while (!queue.isEmpty()) {<a 
name="line.482"></a>
-<span class="sourceLineNo">483</span>      // need to reload split keys each 
iteration.<a name="line.483"></a>
-<span class="sourceLineNo">484</span>      final Pair&lt;byte[][], 
byte[][]&gt; startEndKeys = regionLocator.getStartEndKeys();<a 
name="line.484"></a>
-<span class="sourceLineNo">485</span>      if (count != 0) {<a 
name="line.485"></a>
-<span class="sourceLineNo">486</span>        LOG.info("Split occurred while 
grouping HFiles, retry attempt " +<a name="line.486"></a>
-<span class="sourceLineNo">487</span>            + count + " with " + 
queue.size() + " files remaining to group or split");<a name="line.487"></a>
-<span class="sourceLineNo">488</span>      }<a name="line.488"></a>
-<span class="sourceLineNo">489</span><a name="line.489"></a>
-<span class="sourceLineNo">490</span>      int maxRetries = 
getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a 
name="line.490"></a>
-<span class="sourceLineNo">491</span>      maxRetries = Math.max(maxRetries, 
startEndKeys.getFirst().length + 1);<a name="line.491"></a>
-<span class="sourceLineNo">492</span>      if (maxRetries != 0 &amp;&amp; 
count &gt;= maxRetries) {<a name="line.492"></a>
-<span class="sourceLineNo">493</span>        throw new IOException("Retry 
attempted " + count +<a name="line.493"></a>
-<span class="sourceLineNo">494</span>            " times without completing, 
bailing out");<a name="line.494"></a>
-<span class="sourceLineNo">495</span>      }<a name="line.495"></a>
-<span class="sourceLineNo">496</span>      count++;<a name="line.496"></a>
-<span class="sourceLineNo">497</span><a name="line.497"></a>
-<span class="sourceLineNo">498</span>      // Using ByteBuffer for byte[] 
equality semantics<a name="line.498"></a>
-<span class="sourceLineNo">499</span>      pair = groupOrSplitPhase(table, 
pool, queue, startEndKeys);<a name="line.499"></a>
-<span class="sourceLineNo">500</span>      Multimap&lt;ByteBuffer, 
LoadQueueItem&gt; regionGroups = pair.getFirst();<a name="line.500"></a>
-<span class="sourceLineNo">501</span><a name="line.501"></a>
-<span class="sourceLineNo">502</span>      if 
(!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.502"></a>
-<span class="sourceLineNo">503</span>        // Error is logged inside 
checkHFilesCountPerRegionPerFamily.<a name="line.503"></a>
-<span class="sourceLineNo">504</span>        throw new IOException("Trying to 
load more than " + maxFilesPerRegionPerFamily<a name="line.504"></a>
-<span class="sourceLineNo">505</span>            + " hfiles to one family of 
one region");<a name="line.505"></a>
-<span class="sourceLineNo">506</span>      }<a name="line.506"></a>
-<span class="sourceLineNo">507</span><a name="line.507"></a>
-<span class="sourceLineNo">508</span>      bulkLoadPhase(table, 
admin.getConnection(), pool, queue, regionGroups, copyFile,<a 
name="line.508"></a>
-<span class="sourceLineNo">509</span>          item2RegionMap);<a 
name="line.509"></a>
-<span class="sourceLineNo">510</span><a name="line.510"></a>
-<span class="sourceLineNo">511</span>      // NOTE: The next iteration's split 
/ group could happen in parallel to<a name="line.511"></a>
-<span class="sourceLineNo">512</span>      // atomic bulkloads assuming that 
there are splits and no merges, and<a name="line.512"></a>
-<span class="sourceLineNo">513</span>      // that we can atomically pull out 
the groups we want to retry.<a name="line.513"></a>
-<span class="sourceLineNo">514</span>    }<a name="line.514"></a>
-<span class="sourceLineNo">515</span><a name="line.515"></a>
-<span class="sourceLineNo">516</span>    if (!queue.isEmpty()) {<a 
name="line.516"></a>
-<span class="sourceLineNo">517</span>      throw new RuntimeException("Bulk 
load aborted with some files not yet loaded."<a name="line.517"></a>
-<span class="sourceLineNo">518</span>        + "Please check log for more 
details.");<a name="line.518"></a>
-<span class="sourceLineNo">519</span>    }<a name="line.519"></a>
-<span class="sourceLineNo">520</span>    return item2RegionMap;<a 
name="line.520"></a>
-<span class="sourceLineNo">521</span>  }<a name="line.521"></a>
-<span class="sourceLineNo">522</span><a name="line.522"></a>
-<span class="sourceLineNo">523</span>  protected 
ClientServiceCallable&lt;byte[]&gt; buildClientServiceCallable(final Connection 
conn,<a name="line.523"></a>
-<span class="sourceLineNo">524</span>      TableName tableName, byte[] first, 
Collection&lt;LoadQueueItem&gt; lqis, boolean copyFile) {<a name="line.524"></a>
-<span class="sourceLineNo">525</span><a name="line.525"></a>
-<span class="sourceLineNo">526</span>    final List&lt;Pair&lt;byte[], 
String&gt;&gt; famPaths = new ArrayList&lt;&gt;(lqis.size());<a 
name="line.526"></a>
-<span class="sourceLineNo">527</span>    for (LoadQueueItem lqi : lqis) {<a 
name="line.527"></a>
-<span class="sourceLineNo">528</span>        
famPaths.add(Pair.newPair(lqi.family, lqi.hfilePath.toString()));<a 
name="line.528"></a>
-<span class="sourceLineNo">529</span>    }<a name="line.529"></a>
-<span class="sourceLineNo">530</span><a name="line.530"></a>
-<span class="sourceLineNo">531</span>    return new 
ClientServiceCallable&lt;byte[]&gt;(conn,<a name="line.531"></a>
-<span class="sourceLineNo">532</span>        tableName, first, 
rpcControllerFactory.newController()) {<a name="line.532"></a>
-<span class="sourceLineNo">533</span>      @Override<a name="line.533"></a>
-<span class="sourceLineNo">534</span>      protected byte[] rpcCall() throws 
Exception {<a name="line.534"></a>
-<span class="sourceLineNo">535</span>        SecureBulkLoadClient secureClient 
= null;<a name="line.535"></a>
-<span class="sourceLineNo">536</span>        boolean success = false;<a 
name="line.536"></a>
-<span class="sourceLineNo">537</span>        try {<a name="line.537"></a>
-<span class="sourceLineNo">538</span>          LOG.debug("Going to connect to 
server " + getLocation() + " for row "<a name="line.538"></a>
-<span class="sourceLineNo">539</span>              + 
Bytes.toStringBinary(getRow()) + " with hfile group " +<a name="line.539"></a>
-<span class="sourceLineNo">540</span>              
LoadIncrementalHFiles.this.toString( famPaths));<a name="line.540"></a>
-<span class="sourceLineNo">541</span>          byte[] regionName = 
getLocation().getRegionInfo().getRegionName();<a name="line.541"></a>
-<span class="sourceLineNo">542</span>          try (Table table = 
conn.getTable(getTableName())) {<a name="line.542"></a>
-<span class="sourceLineNo">543</span>            secureClient = new 
SecureBulkLoadClient(getConf(), table);<a name="line.543"></a>
-<span class="sourceLineNo">544</span>            success = 
secureClient.secureBulkLoadHFiles(getStub(), famPaths, regionName,<a 
name="line.544"></a>
-<span class="sourceLineNo">545</span>                assignSeqIds, 
fsDelegationToken.getUserToken(), bulkToken, copyFile);<a name="line.545"></a>
-<span class="sourceLineNo">546</span>          }<a name="line.546"></a>
-<span class="sourceLineNo">547</span>          return success ? regionName : 
null;<a name="line.547"></a>
-<span class="sourceLineNo">548</span>        } finally {<a name="line.548"></a>
-<span class="sourceLineNo">549</span>          //Best effort copying of files 
that might not have been imported<a name="line.549"></a>
-<span class="sourceLineNo">550</span>          //from the staging directory 
back to original location<a name="line.550"></a>
-<span class="sourceLineNo">551</span>          //in user directory<a 
name="line.551"></a>
-<span class="sourceLineNo">552</span>          if (secureClient != null 
&amp;&amp; !success) {<a name="line.552"></a>
-<span class="sourceLineNo">553</span>            FileSystem targetFs = 
FileSystem.get(getConf());<a name="line.553"></a>
-<span class="sourceLineNo">554</span>            // fs is the source 
filesystem<a name="line.554"></a>
-<span class="sourceLineNo">555</span>            if (fs == null) {<a 
name="line.555"></a>
-<span class="sourceLineNo">556</span>              fs = 
lqis.iterator().next().hfilePath.getFileSystem(getConf());<a 
name="line.556"></a>
-<span class="sourceLineNo">557</span>            }<a name="line.557"></a>
-<span class="sourceLineNo">558</span>            // Check to see if the source 
and target filesystems are the same<a name="line.558"></a>
-<span class="sourceLineNo">559</span>            // If they are the same 
filesystem, we will try move the files back<a name="line.559"></a>
-<span class="sourceLineNo">560</span>            // because previously we 
moved them to the staging directory.<a name="line.560"></a>
-<span class="sourceLineNo">561</span>            if 
(FSHDFSUtils.isSameHdfs(getConf(), fs, targetFs)) {<a name="line.561"></a>
-<span class="sourceLineNo">562</span>              for (Pair&lt;byte[], 
String&gt; el : famPaths) {<a name="line.562"></a>
-<span class="sourceLineNo">563</span>                Path hfileStagingPath = 
null;<a name="line.563"></a>
-<span class="sourceLineNo">564</span>                Path hfileOrigPath = new 
Path(el.getSecond());<a name="line.564"></a>
-<span class="sourceLineNo">565</span>                try {<a 
name="line.565"></a>
-<span class="sourceLineNo">566</span>                  hfileStagingPath = new 
Path(new Path(bulkToken, Bytes.toString(el.getFirst())),<a name="line.566"></a>
-<span class="sourceLineNo">567</span>                      
hfileOrigPath.getName());<a name="line.567"></a>
-<span class="sourceLineNo">568</span>                  if 
(targetFs.rename(hfileStagingPath, hfileOrigPath)) {<a name="line.568"></a>
-<span class="sourceLineNo">569</span>                    LOG.debug("Moved back 
file " + hfileOrigPath + " from " +<a name="line.569"></a>
-<span class="sourceLineNo">570</span>                        
hfileStagingPath);<a name="line.570"></a>
-<span class="sourceLineNo">571</span>                  } else if 
(targetFs.exists(hfileStagingPath)) {<a name="line.571"></a>
-<span class="sourceLineNo">572</span>                    LOG.debug("Unable to 
move back file " + hfileOrigPath + " from " +<a name="line.572"></a>
-<span class="sourceLineNo">573</span>                        
hfileStagingPath);<a name="line.573"></a>
-<span class="sourceLineNo">574</span>                  }<a name="line.574"></a>
-<span class="sourceLineNo">575</span>                } catch (Exception ex) 
{<a name="line.575"></a>
-<span class="sourceLineNo">576</span>                  LOG.debug("Unable to 
move back file " + hfileOrigPath + " from " +<a name="line.576"></a>
-<span class="sourceLineNo">577</span>                      hfileStagingPath, 
ex);<a name="line.577"></a>
-<span class="sourceLineNo">578</span>                }<a name="line.578"></a>
-<span class="sourceLineNo">579</span>              }<a name="line.579"></a>
-<span class="sourceLineNo">580</span>            }<a name="line.580"></a>
-<span class="sourceLineNo">581</span>          }<a name="line.581"></a>
-<span class="sourceLineNo">582</span>        }<a name="line.582"></a>
-<span class="sourceLineNo">583</span>      }<a name="line.583"></a>
-<span class="sourceLineNo">584</span>    };<a name="line.584"></a>
-<span class="sourceLineNo">585</span>  }<a name="line.585"></a>
-<span class="sourceLineNo">586</span><a name="line.586"></a>
-<span class="sourceLineNo">587</span>  /**<a name="line.587"></a>
-<span class="sourceLineNo">588</span>   * Prepare a collection of {@link 
LoadQueueItem} from list of source hfiles contained in the<a 
name="line.588"></a>
-<span class="sourceLineNo">589</span>   * passed directory and validates 
whether the prepared queue has all the valid table column<a name="line.589"></a>
-<span class="sourceLineNo">590</span>   * families in it.<a 
name="line.590"></a>
-<span class="sourceLineNo">591</span>   * @param hfilesDir directory 
containing list of hfiles to be loaded into the table<a name="line.591"></a>
-<span class="sourceLineNo">592</span>   * @param table table to which hfiles 
should be loaded<a name="line.592"></a>
-<span class="sourceLineNo">593</span>   * @param queue queue which needs to be 
loaded into the table<a name="line.593"></a>
-<span class="sourceLineNo">594</span>   * @param validateHFile if true hfiles 
will be validated for its format<a name="line.594"></a>
-<span class="sourceLineNo">595</span>   * @throws IOException If any I/O or 
network error occurred<a name="line.595"></a>
-<span class="sourceLineNo">596</span>   */<a name="line.596"></a>
-<span class="sourceLineNo">597</span>  public void prepareHFileQueue(Path 
hfilesDir, Table table, Deque&lt;LoadQueueItem&gt; queue,<a name="line.597"></a>
-<span class="sourceLineNo">598</span>      boolean validateHFile) throws 
IOException {<a name="line.598"></a>
-<span class="sourceLineNo">599</span>    prepareHFileQueue(hfilesDir, table, 
queue, validateHFile, false);<a name="line.599"></a>
-<span class="sourceLineNo">600</span>  }<a name="line.600"></a>
-<span class="sourceLineNo">601</span><a name="line.601"></a>
-<span class="sourceLineNo">602</span>  /**<a name="line.602"></a>
-<span class="sourceLineNo">603</span>   * Prepare a collection of {@link 
LoadQueueItem} from list of source hfiles contained in the<a 
name="line.603"></a>
-<span class="sourceLineNo">604</span>   * passed directory and validates 
whether the prepared queue has all the valid table column<a name="line.604"></a>
-<span class="sourceLineNo">605</span>   * families in it.<a 
name="line.605"></a>
-<span class="sourceLineNo">606</span>   * @param hfilesDir directory 
containing list of hfiles to be loaded into the table<a name="line.606"></a>
-<span class="sourceLineNo">607</span>   * @param table table to which hfiles 
should be loaded<a name="line.607"></a>
-<span class="sourceLineNo">608</span>   * @param queue queue which needs to be 
loaded into the table<a name="line.608"></a>
-<span class="sourceLineNo">609</span>   * @param validateHFile if true hfiles 
will be validated for its format<a name="line.609"></a>
-<span class="sourceLineNo">610</span>   * @param silence  true to ignore 
unmatched column families<a name="line.610"></a>
-<span class="sourceLineNo">611</span>   * @throws IOException If any I/O or 
network error occurred<a name="line.611"></a>
-<span class="sourceLineNo">612</span>   */<a name="line.612"></a>
-<span class="sourceLineNo">613</span>  public void prepareHFileQueue(Path 
hfilesDir, Table table,<a name="line.613"></a>
-<span class="sourceLineNo">614</span>      Deque&lt;LoadQueueItem&gt; queue, 
boolean validateHFile, boolean silence) throws IOException {<a 
name="line.614"></a>
-<span class="sourceLineNo">615</span>    discoverLoadQueue(queue, hfilesDir, 
validateHFile);<a name="line.615"></a>
-<span class="sourceLineNo">616</span>    validateFamiliesInHFiles(table, 
queue, silence);<a name="line.616"></a>
-<span class="sourceLineNo">617</span>  }<a name="line.617"></a>
-<span class="sourceLineNo">618</span><a name="line.618"></a>
-<span class="sourceLineNo">619</span>  /**<a name="line.619"></a>
-<span class="sourceLineNo">620</span>   * Prepare a collection of {@link 
LoadQueueItem} from list of source hfiles contained in the<a 
name="line.620"></a>
-<span class="sourceLineNo">621</span>   * passed directory and validates 
whether the prepared queue has all the valid table column<a name="line.621"></a>
-<span class="sourceLineNo">622</span>   * families in it.<a 
name="line.622"></a>
-<span class="sourceLineNo">623</span>   * @param map map of family to List of 
hfiles<a name="line.623"></a>
-<span class="sourceLineNo">624</span>   * @param table table to which hfiles 
should be loaded<a name="line.624"></a>
-<span class="sourceLineNo">625</span>   * @param queue queue which needs to be 
loaded into the table<a name="line.625"></a>
-<span class="sourceLineNo">626</span>   * @param silence  true to ignore 
unmatched column families<a name="line.626"></a>
-<span class="sourceLineNo">627</span>   * @throws IOException If any I/O or 
network error occurred<a name="line.627"></a>
-<span class="sourceLineNo">628</span>   */<a name="line.628"></a>
-<span class="sourceLineNo">629</span>  public void 
prepareHFileQueue(Map&lt;byte[], List&lt;Path&gt;&gt; map, Table table,<a 
name="line.629"></a>
-<span class="sourceLineNo">630</span>      Deque&lt;LoadQueueItem&gt; queue, 
boolean silence) throws IOException {<a name="line.630"></a>
-<span class="sourceLineNo">631</span>    populateLoadQueue(queue, map);<a 
name="line.631"></a>
-<span class="sourceLineNo">632</span>    validateFamiliesInHFiles(table, 
queue, silence);<a name="line.632"></a>
-<span class="sourceLineNo">633</span>  }<a name="line.633"></a>
-<span class="sourceLineNo">634</span><a name="line.634"></a>
-<span class="sourceLineNo">635</span>  // Initialize a thread pool<a 
name="line.635"></a>
-<span class="sourceLineNo">636</span>  private ExecutorService 
createExecutorService() {<a name="line.636"></a>
-<span class="sourceLineNo">637</span>    ThreadFactoryBuilder builder = new 
ThreadFactoryBuilder();<a name="line.637"></a>
-<span class="sourceLineNo">638</span>    
builder.setNameFormat("LoadIncrementalHFiles-%1$d");<a name="line.638"></a>
-<span class="sourceLineNo">639</span>    ExecutorService pool = new 
ThreadPoolExecutor(nrThreads, nrThreads, 60, TimeUnit.SECONDS,<a 
name="line.639"></a>
-<span class="sourceLineNo">640</span>        new 
LinkedBlockingQueue&lt;&gt;(), builder.build());<a name="line.640"></a>
-<span class="sourceLineNo">641</span>    ((ThreadPoolExecutor) 
pool).allowCoreThreadTimeOut(true);<a name="line.641"></a>
-<span class="sourceLineNo">642</span>    return pool;<a name="line.642"></a>
-<span class="sourceLineNo">643</span>  }<a name="line.643"></a>
-<span class="sourceLineNo">644</span><a name="line.644"></a>
-<span class="sourceLineNo">645</span>  /**<a name="line.645"></a>
-<span class="sourceLineNo">646</span>   * Checks whether there is any invalid 
family name in HFiles to be bulk loaded.<a name="line.646"></a>
-<span class="sourceLineNo">647</span>   */<a name="line.647"></a>
-<span class="sourceLineNo">648</span>  private void 
validateFamiliesInHFiles(Table table, Deque&lt;LoadQueueItem&gt; queue, boolean 
silence)<a name="line.648"></a>
-<span class="sourceLineNo">649</span>      throws IOException {<a 
name="line.649"></a>
-<span class="sourceLineNo">650</span>    Collection&lt;HColumnDescriptor&gt; 
families = table.getTableDescriptor().getFamilies();<a name="line.650"></a>
-<span class="sourceLineNo">651</span>    List&lt;String&gt; familyNames = new 
ArrayList&lt;&gt;(families.size());<a name="line.651"></a>
-<span class="sourceLineNo">652</span>    for (HColumnDescriptor family : 
families) {<a name="line.652"></a>
-<span class="sourceLineNo">653</span>      
familyNames.add(family.getNameAsString());<a name="line.653"></a>
-<span class="sourceLineNo">654</span>    }<a name="line.654"></a>
-<span class="sourceLineNo">655</span>    Iterator&lt;LoadQueueItem&gt; 
queueIter = queue.iterator();<a name="line.655"></a>
-<span class="sourceLineNo">656</span>    while (queueIter.hasNext()) {<a 
name="line.656"></a>
-<span class="sourceLineNo">657</span>      LoadQueueItem lqi = 
queueIter.next();<a name="line.657"></a>
-<span class="sourceLineNo">658</span>      String familyNameInHFile = 
Bytes.toString(lqi.family);<a name="line.658"></a>
-<span class="sourceLineNo">659</span>      if 
(!familyNames.contains(familyNameInHFile)) {<a name="line.659"></a>
-<span class="sourceLineNo">660</span>        
unmatchedFamilies.add(familyNameInHFile);<a name="line.660"></a>
-<span class="sourceLineNo">661</span>      }<a name="line.661"></a>
-<span class="sourceLineNo">662</span>    }<a name="line.662"></a>
-<span class="sourceLineNo">663</span>    if (unmatchedFamilies.size() &gt; 0) 
{<a name="line.663"></a>
-<span class="sourceLineNo">664</span>      String msg =<a name="line.664"></a>
-<span class="sourceLineNo">665</span>          "Unmatched family names found: 
unmatched family names in HFiles to be bulkloaded: "<a name="line.665"></a>
-<span class="sourceLineNo">666</span>              + unmatchedFamilies + "; 
valid family names of table " + table.getName() + " are: "<a 
name="line.666"></a>
-<span class="sourceLineNo">667</span>              + familyNames;<a 
name="line.667"></a>
-<span class="sourceLineNo">668</span>      LOG.error(msg);<a 
name="line.668"></a>
-<span class="sourceLineNo">669</span>      if (!silence) throw new 
IOException(msg);<a name="line.669"></a>
-<span class="sourceLineNo">670</span>    }<a name="line.670"></a>
-<span class="sourceLineNo">671</span>  }<a name="line.671"></a>
-<span class="sourceLineNo">672</span><a name="line.672"></a>
-<span class="sourceLineNo">673</span>  /**<a name="line.673"></a>
-<span class="sourceLineNo">674</span>   * Used by the replication sink to load 
the hfiles from the source cluster. It does the following,<a 
name="line.674"></a>
-<span class="sourceLineNo">675</span>   * &lt;ol&gt;<a name="line.675"></a>
-<span class="sourceLineNo">676</span>   * 
&lt;li&gt;LoadIncrementalHFiles#groupOrSplitPhase(Table, ExecutorService, 
Deque, Pair)}&lt;/li&gt;<a name="line.676"></a>
-<span class="sourceLineNo">677</span>   * 
&lt;li&gt;LoadIncrementalHFiles#bulkLoadPhase(Table, Connection, 
ExecutorService, Deque, Multimap)<a name="line.677"></a>
-<span class="sourceLineNo">678</span>   * &lt;/li&gt;<a name="line.678"></a>
-<span class="sourceLineNo">679</span>   * &lt;/ol&gt;<a name="line.679"></a>
-<span class="sourceLineNo">680</span>   * @param table Table to which these 
hfiles should be loaded to<a name="line.680"></a>
-<span class="sourceLineNo">681</span>   * @param conn Connection to use<a 
name="line.681"></a>
-<span class="sourceLineNo">682</span>   * @param queue {@link LoadQueueItem} 
has hfiles yet to be loaded<a name="line.682"></a>
-<span class="sourceLineNo">683</span>   * @param startEndKeys starting and 
ending row keys of the region<a name="line.683"></a>
-<span class="sourceLineNo">684</span>   */<a name="line.684"></a>
-<span class="sourceLineNo">685</span>  public void loadHFileQueue(final Table 
table, final Connection conn, Deque&lt;LoadQueueItem&gt; queue,<a 
name="line.685"></a>
-<span class="sourceLineNo">686</span>      Pair&lt;byte[][], byte[][]&gt; 
startEndKeys) throws IOException {<a name="line.686"></a>
-<span class="sourceLineNo">687</span>    loadHFileQueue(table, conn, queue, 
startEndKeys, false);<a name="line.687"></a>
-<span class="sourceLineNo">688</span>  }<a name="line.688"></a>
-<span class="sourceLineNo">689</span><a name="line.689"></a>
-<span class="sourceLineNo">690</span>  /**<a name="line.690"></a>
-<span class="sourceLineNo">691</span>   * Used by the replication sink to load 
the hfiles from the source cluster. It does the following,<a 
name="line.691"></a>
-<span class="sourceLineNo">692</span>   * &lt;ol&gt;<a name="line.692"></a>
-<span class="sourceLineNo">693</span>   * 
&lt;li&gt;LoadIncrementalHFiles#groupOrSplitPhase(Table, ExecutorService, 
Deque, Pair)}&lt;/li&gt;<a name="line.693"></a>
-<span class="sourceLineNo">694</span>   * 
&lt;li&gt;LoadIncrementalHFiles#bulkLoadPhase(Table, Connection, 
ExecutorService, Deque, Multimap)<a name="line.694"></a>
-<span class="sourceLineNo">695</span>   * &lt;/li&gt;<a name="line.695"></a>
-<span class="sourceLineNo">696</span>   * &lt;/ol&gt;<a name="line.696"></a>
-<span class="sourceLineNo">697</span>   * @param table Table to which these 
hfiles should be loaded to<a name="line.697"></a>
-<span class="sourceLineNo">698</span>   * @param conn Connection to use<a 
name="line.698"></a>
-<span class="sourceLineNo">699</span>   * @param queue {@link LoadQueueItem} 
has hfiles yet to be loaded<a name="line.699"></a>
-<span class="sourceLineNo">700</span>   * @param startEndKeys starting and 
ending row keys of the region<a name="line.700"></a>
-<span class="sourceLineNo">701</span>   */<a name="line.701"></a>
-<span class="sourceLineNo">702</span>  public void loadHFileQueue(final Table 
table, final Connection conn, Deque&lt;LoadQueueItem&gt; queue,<a 
name="line.702"></a>
-<span class="sourceLineNo">703</span>      Pair&lt;byte[][], byte[][]&gt; 
startEndKeys, boolean copyFile) throws IOException {<a name="line.703"></a>
-<span class="sourceLineNo">704</span>    ExecutorService pool = null;<a 
name="line.704"></a>
-<span class="sourceLineNo">705</span>    try {<a name="line.705"></a>
-<span class="sourceLineNo">706</span>      pool = createExecutorService();<a 
name="line.706"></a>
-<span class="sourceLineNo">707</span>      Multimap&lt;ByteBuffer, 
LoadQueueItem&gt; regionGroups =<a name="line.707"></a>
-<span class="sourceLineNo">708</span>          groupOrSplitPhase(table, pool, 
queue, startEndKeys).getFirst();<a name="line.708"></a>
-<span class="sourceLineNo">709</span>      bulkLoadPhase(table, conn, pool, 
queue, regionGroups, copyFile, null);<a name="line.709"></a>
-<span class="sourceLineNo">710</span>    } finally {<a name="line.710"></a>
-<span class="sourceLineNo">711</span>      if (pool != null) {<a 
name="line.711"></a>
-<span class="sourceLineNo">712</span>        pool.shutdown();<a 
name="line.712"></a>
-<span class="sourceLineNo">713</span>      }<a name="line.713"></a>
-<span class="sourceLineNo">714</span>    }<a name="line.714"></a>
-<span class="sourceLineNo">715</span>  }<a name="line.715"></a>
-<span class="sourceLineNo">716</span><a name="line.716"></a>
-<span class="sourceLineNo">717</span>  /**<a name="line.717"></a>
-<span class="sourceLineNo">718</span>   * This takes the LQI's grouped by 
likely regions and attempts to bulk load<a name="line.718"></a>
-<span class="sourceLineNo">719</span>   * them.  Any failures are re-queued 
for another pass with the<a name="line.719"></a>
-<span class="sourceLineNo">720</span>   * groupOrSplitPhase.<a 
name="line.720"></a>
-<span class="sourceLineNo">721</span>   */<a name="line.721"></a>
-<span class="sourceLineNo">722</span>  protected void bulkLoadPhase(final 
Table table, final Connection conn,<a name="line.722"></a>
-<span class="sourceLineNo">723</span>      ExecutorService pool, 
Deque&lt;LoadQueueItem&gt; queue,<a name="line.723"></a>
-<span class="sourceLineNo">724</span>      final Multimap&lt;ByteBuffer, 
LoadQueueItem&gt; regionGroups, boolean copyFile,<a name="line.724"></a>
-<span class="sourceLineNo">725</span>      Map&lt;LoadQueueItem, 
ByteBuffer&gt; item2RegionMap) throws IOException {<a name="line.725"></a>
-<span class="sourceLineNo">726</span>    // atomically bulk load the groups.<a 
name="line.726"></a>
-<span class="sourceLineNo">727</span>    
Set&lt;Future&lt;List&lt;LoadQueueItem&gt;&gt;&gt; loadingFutures = new 
HashSet&lt;&gt;();<a name="line.727"></a>
-<span class="sourceLineNo">728</span>    for (Entry&lt;ByteBuffer, ? extends 
Collection&lt;LoadQueueItem&gt;&gt; e: regionGroups.asMap().entrySet()){<a 
name="line.728"></a>
-<span class="sourceLineNo">729</span>      final byte[] first = 
e.getKey().array();<a name="line.729"></a>
-<span class="sourceLineNo">730</span>      final 
Collection&lt;LoadQueueItem&gt; lqis =  e.getValue();<a name="line.730"></a>
-<span class="sourceLineNo">731</span><a name="line.731"></a>
-<span class="sourceLineNo">732</span>      final 
ClientServiceCallable&lt;byte[]&gt; serviceCallable =<a name="line.732"></a>
-<span class="sourceLineNo">733</span>          
buildClientServiceCallable(conn, table.getName(), first, lqis, copyFile);<a 
name="line.733"></a>
-<span class="sourceLineNo">734</span><a name="line.734"></a>
-<span class="sourceLineNo">735</span>      final 
Callable&lt;List&lt;LoadQueueItem&gt;&gt; call = new 
Callable&lt;List&lt;LoadQueueItem&gt;&gt;() {<a name="line.735"></a>
-<span class="sourceLineNo">736</span>        @Override<a name="line.736"></a>
-<span class="sourceLineNo">737</span>        public List&lt;LoadQueueItem&gt; 
call() throws Exception {<a name="line.737"></a>
-<span class="sourceLineNo">738</span>          List&lt;LoadQueueItem&gt; 
toRetry =<a name="line.738"></a>
-<span class="sourceLineNo">739</span>              
tryAtomicRegionLoad(serviceCallable, table.getName(), first, lqis);<a 
name="line.739"></a>
-<span class="sourceLineNo">740</span>          return toRetry;<a 
name="line.740"></a>
-<span class="sourceLineNo">741</span>        }<a name="line.741"></a>
-<span class="sourceLineNo">742</span>      };<a name="line.742"></a>
-<span class="sourceLineNo">743</span>      if (item2RegionMap != null) {<a 
name="line.743"></a>
-<span class="sourceLineNo">744</span>        for (LoadQueueItem lqi : lqis) 
{<a name="line.744"></a>
-<span class="sourceLineNo">745</span>          item2RegionMap.put(lqi, 
e.getKey());<a name="line.745"></a>
-<span class="sourceLineNo">746</span>        }<a name="line.746"></a>
-<span class="sourceLineNo">747</span>      }<a name="line.747"></a>
-<span class="sourceLineNo">748</span>      
loadingFutures.add(pool.submit(call));<a name="line.748"></a>
-<span class="sourceLineNo">749</span>    }<a name="line.749"></a>
-<span class="sourceLineNo">750</span><a name="line.750"></a>
-<span class="sourceLineNo">751</span>    // get all the results.<a 
name="line.751"></a>
-<span class="sourceLineNo">752</span>    for 
(Future&lt;List&lt;LoadQueueItem&gt;&gt; future : loadingFutures) {<a 
name="line.752"></a>
-<span class="sourceLineNo">753</span>      try {<a name="line.753"></a>
-<span class="sourceLineNo">754</span>        List&lt;LoadQueueItem&gt; toRetry 
= future.get();<a name="line.754"></a>
-<span class="sourceLineNo">755</span><a name="line.755"></a>
-<span class="sourceLineNo">756</span>        if (item2RegionMap != null) {<a 
name="line.756"></a>
-<span class="sourceLineNo">757</span>          for (LoadQueueItem lqi : 
toRetry) {<a name="line.757"></a>
-<span class="sourceLineNo">758</span>            item2RegionMap.remove(lqi);<a 
name="line.758"></a>
-<span class="sourceLineNo">759</span>          }<a name="line.759"></a>
-<span class="sourceLineNo">760</span>        }<a name="line.760"></a>
-<span class="sourceLineNo">761</span>        // LQIs that are requeued to be 
regrouped.<a name="line.761"></a>
-<span class="sourceLineNo">762</span>        queue.addAll(toRetry);<a 
name="line.762"></a>
-<span class="sourceLineNo">763</span><a name="line.763"></a>
-<span class="sourceLineNo">764</span>      } catch (ExecutionException e1) {<a 
name="line.764"></a>
-<span class="sourceLineNo">765</span>        Throwable t = e1.getCause();<a 
name="line.765"></a>
-<span class="sourceLineNo">766</span>        if (t instanceof IOException) {<a 
name="line.766"></a>
-<span class="sourceLineNo">767</span>          // At this point something 
unrecoverable has happened.<a name="line.767"></a>
-<span class="sourceLineNo">768</span>          // TODO Implement bulk load 
recovery<a name="line.768"></a>
-<span class="sourceLineNo">769</span>          throw new IOException("BulkLoad 
encountered an unrecoverable problem", t);<a name="line.769"></a>
-<span class="sourceLineNo">770</span>        }<a name="line.770"></a>
-<span class="sourceLineNo">771</span>        LOG.error("Unexpected execution 
exception during bulk load", e1);<a name="line.771"></a>
-<span class="sourceLineNo">772</span>        throw new 
IllegalStateException(t);<a name="line.772"></a>
-<span class="sourceLineNo">773</span>      } catch (InterruptedException e1) 
{<a name="line.773"></a>
-<span class="sourceLineNo">774</span>        LOG.error("Unexpected interrupted 
exception during bulk load", e1);<a name="line.774"></a>
-<span class="sourceLineNo">775</span>        throw (InterruptedIOException)new 
InterruptedIOException().initCause(e1);<a name="line.775"></a>
-<span class="sourceLineNo">776</span>      }<a name="line.776"></a>
-<span class="sourceLineNo">777</span>    }<a name="line.777"></a>
-<span class="sourceLineNo">778</span>  }<a name="line.778"></a>
-<span class="sourceLineNo">779</span><a name="line.779"></a>
-<span class="sourceLineNo">780</span>  private boolean 
checkHFilesCountPerRegionPerFamily(<a name="line.780"></a>
-<span class="sourceLineNo">781</span>      final Multimap&lt;ByteBuffer, 
LoadQueueItem&gt; regionGroups) {<a name="line.781"></a>
-<span class="sourceLineNo">782</span>    for (Entry&lt;ByteBuffer,<a 
name="line.782"></a>
-<span class="sourceLineNo">783</span>      ? extends 
Collection&lt;LoadQueueItem&gt;&gt; e: regionGroups.asMap().entrySet()) {<a 
name="line.783"></a>
-<span class="sourceLineNo">784</span>      final 
Collection&lt;LoadQueueItem&gt; lqis =  e.getValue();<a name="line.784"></a>
-<span class="sourceLineNo">785</span>      HashMap&lt;byte[], MutableInt&gt; 
filesMap = new HashMap&lt;&gt;();<a name="line.785"></a>
-<span class="sourceLineNo">786</span>      for (LoadQueueItem lqi: lqis) {<a 
name="line.786"></a>
-<span class="sourceLineNo">787</span>        MutableInt count = 
filesMap.get(lqi.family);<a name="line.787"></a>
-<span class="sourceLineNo">788</span>        if (count == null) {<a 
name="line.788"></a>
-<span class="sourceLineNo">789</span>          count = new MutableInt();<a 
name="line.789"></a>
-<span class="sourceLineNo">790</span>          filesMap.put(lqi.family, 
count);<a name="line.790"></a>
-<span class="sourceLineNo">791</span>        }<a name="line.791"></a>
-<span class="sourceLineNo">792</span>        count.increment();<a 
name="line.792"></a>
-<span class="sourceLineNo">793</span>        if (count.intValue() &gt; 
maxFilesPerRegionPerFamily) {<a name="line.793"></a>
-<span class="sourceLineNo">794</span>          LOG.error("Trying to load more 
than " + maxFilesPerRegionPerFamily<a name="line.794"></a>
-<span class="sourceLineNo">795</span>            + " hfiles to family " + 
Bytes.toStringBinary(lqi.family)<a name="line.795"></a>
-<span class="sourceLineNo">796</span>            + " of region with start key 
"<a name="line.796"></a>
-<span class="sourceLineNo">797</span>            + 
Bytes.toStringBinary(e.getKey()));<a name="line.797"></a>
-<span class="sourceLineNo">798</span>          return false;<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>    return true;<a name="line.802"></a>
-<span class="sourceLineNo">803</span>  }<a name="line.803"></a>
-<span class="sourceLineNo">804</span><a name="line.804"></a>
-<span class="sourceLineNo">805</span>  /**<a name="line.805"></a>
-<span class="sourceLineNo">806</span>   * @param table the table to load 
into<a name="line.806"></a>
-<span class="sourceLineNo">807</span>   * @param pool the ExecutorService<a 
name="line.807"></a>
-<span class="sourceLineNo">808</span>   * @param queue the queue for 
LoadQueueItem<a name="line.808"></a>
-<span class="sourceLineNo">809</span>   * @param startEndKeys start and end 
keys<a name="line.809"></a>
-<span class="sourceLineNo">810</span>   * @return A map that groups LQI by 
likely bulk load region targets and Set of missing hfiles.<a 
name="line.810"></a>
-<span class="sourceLineNo">811</span>   */<a name="line.811"></a>
-<span class="sourceLineNo">812</span>  private Pair&lt;Multimap&lt;ByteBuffer, 
LoadQueueItem&gt;, Set&lt;String&gt;&gt; groupOrSplitPhase(<a 
name="line.812"></a>
-<span class="sourceLineNo">813</span>      final Table table, ExecutorService 
pool, Deque&lt;LoadQueueItem&gt; queue,<a name="line.813"></a>
-<span class="sourceLineNo">814</span>      final Pair&lt;byte[][], 
byte[][]&gt; startEndKeys) throws IOException {<a name="line.814"></a>
-<span class="sourceLineNo">815</span>    // &lt;region start key, LQI&gt; need 
synchronized only within this scope of this<a name="line.815"></a>
-<span class="sourceLineNo">816</span>    // phase because of the puts that 
happen in futures.<a name="line.816"></a>
-<span class="sourceLineNo">817</span>    Multimap&lt;ByteBuffer, 
LoadQueueItem&gt; rgs = HashMultimap.create();<a name="line.817"></a>
-<span class="sourceLineNo">818</span>    final Multimap&lt;ByteBuffer, 
LoadQueueItem&gt; regionGroups = Multimaps.synchronizedMultimap(rgs);<a 
name="line.818"></a>
-<span class="sourceLineNo">819</span>    Set&lt;String&gt; missingHFiles = new 
HashSet&lt;&gt;();<a name="line.819"></a>
-<span class="sourceLineNo">820</span>    Pair&lt;Multimap&lt;ByteBuffer, 
LoadQueueItem&gt;, Set&lt;String&gt;&gt; pair = new 
Pair&lt;&gt;(regionGroups,<a name="line.820"></a>
-<span class="sourceLineNo">821</span>        missingHFiles);<a 
name="line.821"></a>
-<span class="sourceLineNo">822</span><a name="line.822"></a>
-<span class="sourceLineNo">823</span>    // drain LQIs and figure out bulk 
load groups<a name="line.823"></a>
-<span class="sourceLineNo">824</span>    
Set&lt;Future&lt;Pair&lt;List&lt;LoadQueueItem&gt;, String&gt;&gt;&gt; 
splittingFutures = new HashSet&lt;&gt;();<a name="line.824"></a>
-<span class="sourceLineNo">825</span>    while (!queue.isEmpty()) {<a 
name="line.825"></a>
-<span class="sourceLineNo">826</span>      final LoadQueueItem item = 
queue.remove();<a name="line.826"></a>
-<span class="sourceLineNo">827</span><a name="line.827"></a>
-<span class="sourceLineNo">828</span>      final 
Callable&lt;Pair&lt;List&lt;LoadQueueItem&gt;, String&gt;&gt; call =<a 
name="line.828"></a>
-<span class="sourceLineNo">829</span>          new 
Callable&lt;Pair&lt;List&lt;LoadQueueItem&gt;, String&gt;&gt;() {<a 
name="line.829"></a>
-<span class="sourceLineNo">830</span>        @Override<a name="line.830"></a>
-<span class="sourceLineNo">831</span>        public 
Pair&lt;List&lt;LoadQueueItem&gt;, String&gt; call() throws Exception {<a 
name="line.831"></a>
-<span class="sourceLineNo">832</span>          
Pair&lt;List&lt;LoadQueueItem&gt;, String&gt; splits = 
groupOrSplit(regionGroups, item, table,<a name="line.832"></a>
-<span class="sourceLineNo">833</span>              startEndKeys);<a 
name="line.833"></a>
-<span class="sourceLineNo">834</span>          return splits;<a 
name="line.834"></a>
-<span class="sourceLineNo">835</span>        }<a name="line.835"></a>
-<span class="sourceLineNo">836</span>      };<a name="line.836"></a>
-<span class="sourceLineNo">837</span>      
splittingFutures.add(pool.submit(call));<a name="line.837"></a>
-<span class="sourceLineNo">838</span>    }<a name="line.838"></a>
-<span class="sourceLineNo">839</span>    // get all the results.  All grouping 
and splitting must finish before<a name="line.839"></a>
-<span class="sourceLineNo">840</span>    // we can attempt the atomic loads.<a 
name="line.840"></a>
-<span class="sourceLineNo">841</span>    for 
(Future&lt;Pair&lt;List&lt;LoadQueueItem&gt;, String&gt;&gt; lqis : 
splittingFutures) {<a name="line.841"></a>
-<span class="sourceLineNo">842</span>      try {<a name="line.842"></a>
-<span class="sourceLineNo">843</span>        
Pair&lt;List&lt;LoadQueueItem&gt;, String&gt; splits = lqis.get();<a 
name="line.843"></a>
-<span class="sourceLineNo">844</span>        if (splits != null) {<a 
name="line.844"></a>
-<span class="sourceLineNo">845</span>          if (splits.getFirst() != null) 
{<a name="line.845"></a>
-<span class="sourceLineNo">846</span>            
queue.addAll(splits.getFirst());<a name="line.846"></a>
-<span class="sourceLineNo">847</span>          } else {<a name="line.847"></a>
-<span class="sourceLineNo">848</span>            
missingHFiles.add(splits.getSecond());<a name="line.848"></a>
-<span class="sourceLineNo">849</span>          }<a name="line.849"></a>
-<span class="sourceLineNo">850</span>        }<a name="line.850"></a>
-<span class="sourceLineNo">851</span>      } catch (ExecutionException e1) {<a 
name="line.851"></a>
-<span class="sourceLineNo">852</span>        Throwable t = e1.getCause();<a 
name="line.852"></a>
-<span class="sourceLineNo">853</span>        if (t instanceof IOException) {<a 
name="line.853"></a>
-<span class="sourceLineNo">854</span>          LOG.error("IOException during 
splitting", e1);<a name="line.854"></a>
-<span class="sourceLineNo">855</span>          throw (IOException)t; // would 
have been thrown if not parallelized,<a name="line.855"></a>
-<span class="sourceLineNo">856</span>        }<a name="line.856"></a>
-<span class="sourceLineNo">857</span>        LOG.error("Unexpected execution 
exception during splitting", e1);<a name="line.857"></a>
-<span class="sourceLineNo">858</span>        throw new 
IllegalStateException(t);<a name="line.858"></a>
-<span class="sourceLineNo">859</span>      } catch (InterruptedException e1) 
{<a name="line.859"></a>
-<span class="sourceLineNo">860</span>        LOG.error("Unexpected interrupted 
exception during splitting", e1);<a name="line.860"></a>
-<span class="sourceLineNo">861</span>        throw (InterruptedIOException)new 
InterruptedIOException().initCause(e1);<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>    return pair;<a name="line.864"></a>
-<span class="sourceLineNo">865</span>  }<a name="line.865"></a>
-<span class="sourceLineNo">866</span><a name="line.866"></a>
-<span class="sourceLineNo">867</span>  // unique file name for the table<a 
name="line.867"></a>
-<span class="sourceLineNo">868</span>  private String getUniqueName() {<a 
name="line.868"></a>
-<span class="sourceLineNo">869</span>    return 
UUID.randomUUID().toString().replaceAll("-", "");<a name="line.869"></a>
-<span class="sourceLineNo">870</span>  }<a name="line.870"></a>
-<span class="sourceLineNo">871</span><a name="line.871"></a>
-<span class="sourceLineNo">872</span>  protected List&lt;LoadQueueItem&gt; 
splitStoreFile(final LoadQueueItem item,<a name="line.872"></a>
-<span class="sourceLineNo">873</span>      final Table table, byte[] 
startKey,<a name="line.873"></a>
-<span class="sourceLineNo">874</span>      byte[] splitKey) throws IOException 
{<a name="line.874"></a>
-<span class="sourceLineNo">875</span>    final Path hfilePath = 
item.hfilePath;<a name="line.875"></a>
-<span class="sourceLineNo">876</span><a name="line.876"></a>
-<span class="sourceLineNo">877</span>    Path tmpDir = 
item.hfilePath.getParent();<a name="line.877"></a>
-<span class="sourceLineNo">878</span>    if 
(!tmpDir.getName().equals(TMP_DIR)) {<a name="line.878"></a>
-<span class="sourceLineNo">879</span>      tmpDir = new Path(tmpDir, 
TMP_DIR);<a name="line.879"></a>
-<span class="sourceLineNo">880</span>    }<a name="line.880"></a>
-<span class="sourceLineNo">881</span><a name="line.881"></a>
-<span class="sourceLineNo">882</span>    LOG.info("HFile at " + hfilePath + " 
no longer fits inside a single " +<a name="line.882"></a>
-<span class="sourceLineNo">883</span>      "region. Splitting...");<a 
name="line.883"></a>
-<span class="sourceLineNo">884</span><a name="line.884"></a>
-<span class="sourceLineNo">885</span>    String uniqueName = 
getUniqueName();<a name="line.885"></a>
-<span class="sourceLineNo">886</span>    HColumnDescriptor familyDesc = 
table.getTableDescriptor().getFamily(item.family);<a name="line.886"></a>
-<span class="sourceLineNo">887</span><a name="line.887"></a>
-<span class="sourceLineNo">888</span>    Path botOut = new Path(tmpDir, 
uniqueName + ".bottom");<a name="line.888"></a>
-<span class="sourceLineNo">889</span>    Path topOut = new Path(tmpDir, 
uniqueName + ".top");<a name="line.889"></a>
-<span class="sourceLineNo">890</span>    splitStoreFile(getConf(), hfilePath, 
familyDesc, splitKey, botOut, topOut);<a name="line.890"></a>
-<span class="sourceLineNo">891</span><a name="line.891"></a>
-<span class="sourceLineNo">892</span>    FileSystem fs = 
tmpDir.getFileSystem(getConf());<a name="line.892"></a>
-<span class="sourceLineNo">893</span>    fs.setPermission(tmpDir, 
FsPermission.valueOf("-rwxrwxrwx"));<a name="line.893"></a>
-<span class="sourceLineNo">894</span>    fs.setPermission(botOut, 
FsPermission.valueOf("-rwxrwxrwx"));<a name="line.894"></a>
-<span class="sourceLineNo">895</span>    fs.setPermission(topOut, 
FsPermission.valueOf("-rwxrwxrwx"));<a name="line.895"></a>
-<span class="sourceLineNo">896</span><a name="line.896"></a>
-<span class="sourceLineNo">897</span>    // Add these back at the *front* of 
the queue, so there's a lower<a name="line.897"></a>
-<span class="sourceLineNo">898</span>    // chance that the region will just 
split again before we get there.<a name="line.898"></a>
-<span class="sourceLineNo">899</span>    List&lt;LoadQueueItem&gt; lqis = new 
ArrayList&lt;&gt;(2);<a name="line.899"></a>
-<span class="sourceLineNo">900</span>    lqis.add(new 
LoadQueueItem(item.family, botOut));<a name="line.900"></a>
-<span class="sourceLineNo">901</span>    lqis.add(new 
LoadQueueItem(item.family, topOut));<a name="line.901"></a>
-<span class="sourceLineNo">902</span><a name="line.902"></a>
-<span class="sourceLineNo">903</span>    // If the current item is already the 
result of previous splits,<a name="line.903"></a>
-<span class="sourceLineNo">904</span>    // we don't need it anymore. Clean up 
to save space.<a name="line.904"></a>
-<span class="sourceLineNo">905</span>    // It is not part of the original 
input files.<a name="line.905"></a>
-<span class="sourceLineNo">906</span>    try {<a name="line.906"></a>
-<span class="sourceLineNo">907</span>      tmpDir = 
item.hfilePath.getParent();<a name="line.907"></a>
-<span class="sourceLineNo">908</span>      if 
(tmpDir.getName().equals(TMP_DIR)) {<a name="line.908"></a>
-<span class="sourceLineNo">909</span>        fs.delete(item.hfilePath, 
false);<a name="line.909"></a>
-<span class="sourceLineNo">910</span>      }<a name="line.910"></a>
-<span class="sourceLineNo">911</span>    } catch (IOException e) {<a 
name="line.911"></a>
-<span class="sourceLineNo">912</span>      LOG.warn("Unable to delete 
temporary split file " + item.hfilePath);<a name="line.912"></a>
-<span class="sourceLineNo">913</span>    }<a name="line.913"></a>
-<span class="sourceLineNo">914</span>    LOG.info("Successfully split into new 
HFiles " + botOut + " and " + topOut);<a name="line.914"></a>
-<span class="sourceLineNo">915</span>    return lqis;<a name="line.915"></a>
-<span class="sourceLineNo">916</span>  }<a name="line.916"></a>
-<span class="sourceLineNo">917</span><a name="line.917"></a>
-<span class="sourceLineNo">918</span>  /**<a name="line.918"></a>
-<span class="sourceLineNo">919</span>   * Attempt to assign the given load 
queue item into its target region group.<a name="line.919"></a>
-<span class="sourceLineNo">920</span>   * If the hfile boundary no longer fits 
into a region, physically splits<a name="line.920"></a>
-<span class="sourceLineNo">921</span>   * the hfile such that the new bottom 
half will fit and returns the list of<a name="line.921"></a>
-<span class="sourceLineNo">922</span>   * LQI's corresponding to the resultant 
hfiles.<a name="line.922"></a>
-<span class="sourceLineNo">923</span>   *<a name="line.923"></a>
-<span class="sourceLineNo">924</span>   * protected for testing<a 
name="line.924"></a>
-<span class="sourceLi

<TRUNCATED>

Reply via email to