You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by mi...@apache.org on 2016/08/31 18:42:37 UTC
[38/52] [partial] hbase-site git commit: Published site at
45af3831fead400115d32552615ea1f11629b03d.
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/8569bd2f/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 a3f7759..908b8b7 100644
--- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
+++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
@@ -122,980 +122,1020 @@
<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><a name="line.117"></a>
-<span class="sourceLineNo">118</span> // We use a '.' prefix which is ignored when walking directory trees<a name="line.118"></a>
-<span class="sourceLineNo">119</span> // above. It is invalid family name.<a name="line.119"></a>
-<span class="sourceLineNo">120</span> final static String TMP_DIR = ".tmp";<a name="line.120"></a>
-<span class="sourceLineNo">121</span><a name="line.121"></a>
-<span class="sourceLineNo">122</span> private int maxFilesPerRegionPerFamily;<a name="line.122"></a>
-<span class="sourceLineNo">123</span> private boolean assignSeqIds;<a name="line.123"></a>
-<span class="sourceLineNo">124</span><a name="line.124"></a>
-<span class="sourceLineNo">125</span> // Source filesystem<a name="line.125"></a>
-<span class="sourceLineNo">126</span> private FileSystem fs;<a name="line.126"></a>
-<span class="sourceLineNo">127</span> // Source delegation token<a name="line.127"></a>
-<span class="sourceLineNo">128</span> private FsDelegationToken fsDelegationToken;<a name="line.128"></a>
-<span class="sourceLineNo">129</span> private String bulkToken;<a name="line.129"></a>
-<span class="sourceLineNo">130</span> private UserProvider userProvider;<a name="line.130"></a>
-<span class="sourceLineNo">131</span> private int nrThreads;<a name="line.131"></a>
-<span class="sourceLineNo">132</span> private RpcControllerFactory rpcControllerFactory;<a name="line.132"></a>
-<span class="sourceLineNo">133</span><a name="line.133"></a>
-<span class="sourceLineNo">134</span> private LoadIncrementalHFiles() {}<a name="line.134"></a>
+<span class="sourceLineNo">117</span> public final static String SILENCE_CONF_KEY = "ignore.unmatched.families";<a name="line.117"></a>
+<span class="sourceLineNo">118</span><a name="line.118"></a>
+<span class="sourceLineNo">119</span> // We use a '.' prefix which is ignored when walking directory trees<a name="line.119"></a>
+<span class="sourceLineNo">120</span> // above. It is invalid family name.<a name="line.120"></a>
+<span class="sourceLineNo">121</span> final static String TMP_DIR = ".tmp";<a name="line.121"></a>
+<span class="sourceLineNo">122</span><a name="line.122"></a>
+<span class="sourceLineNo">123</span> private int maxFilesPerRegionPerFamily;<a name="line.123"></a>
+<span class="sourceLineNo">124</span> private boolean assignSeqIds;<a name="line.124"></a>
+<span class="sourceLineNo">125</span> private Set<String> unmatchedFamilies = new HashSet<String>();<a name="line.125"></a>
+<span class="sourceLineNo">126</span><a name="line.126"></a>
+<span class="sourceLineNo">127</span> // Source filesystem<a name="line.127"></a>
+<span class="sourceLineNo">128</span> private FileSystem fs;<a name="line.128"></a>
+<span class="sourceLineNo">129</span> // Source delegation token<a name="line.129"></a>
+<span class="sourceLineNo">130</span> private FsDelegationToken fsDelegationToken;<a name="line.130"></a>
+<span class="sourceLineNo">131</span> private String bulkToken;<a name="line.131"></a>
+<span class="sourceLineNo">132</span> private UserProvider userProvider;<a name="line.132"></a>
+<span class="sourceLineNo">133</span> private int nrThreads;<a name="line.133"></a>
+<span class="sourceLineNo">134</span> private RpcControllerFactory rpcControllerFactory;<a name="line.134"></a>
<span class="sourceLineNo">135</span><a name="line.135"></a>
-<span class="sourceLineNo">136</span> public LoadIncrementalHFiles(Configuration conf) throws Exception {<a name="line.136"></a>
-<span class="sourceLineNo">137</span> super(conf);<a name="line.137"></a>
-<span class="sourceLineNo">138</span> this.rpcControllerFactory = new RpcControllerFactory(conf);<a name="line.138"></a>
-<span class="sourceLineNo">139</span> initialize();<a name="line.139"></a>
-<span class="sourceLineNo">140</span> }<a name="line.140"></a>
-<span class="sourceLineNo">141</span><a name="line.141"></a>
-<span class="sourceLineNo">142</span> private void initialize() throws Exception {<a name="line.142"></a>
-<span class="sourceLineNo">143</span> if (initalized) {<a name="line.143"></a>
-<span class="sourceLineNo">144</span> return;<a name="line.144"></a>
-<span class="sourceLineNo">145</span> }<a name="line.145"></a>
-<span class="sourceLineNo">146</span> // make a copy, just to be sure we're not overriding someone else's config<a name="line.146"></a>
-<span class="sourceLineNo">147</span> setConf(HBaseConfiguration.create(getConf()));<a name="line.147"></a>
-<span class="sourceLineNo">148</span> Configuration conf = getConf();<a name="line.148"></a>
-<span class="sourceLineNo">149</span> // disable blockcache for tool invocation, see HBASE-10500<a name="line.149"></a>
-<span class="sourceLineNo">150</span> conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.150"></a>
-<span class="sourceLineNo">151</span> this.userProvider = UserProvider.instantiate(conf);<a name="line.151"></a>
-<span class="sourceLineNo">152</span> this.fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.152"></a>
-<span class="sourceLineNo">153</span> assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.153"></a>
-<span class="sourceLineNo">154</span> maxFilesPerRegionPerFamily = conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.154"></a>
-<span class="sourceLineNo">155</span> nrThreads = conf.getInt("hbase.loadincremental.threads.max",<a name="line.155"></a>
-<span class="sourceLineNo">156</span> Runtime.getRuntime().availableProcessors());<a name="line.156"></a>
-<span class="sourceLineNo">157</span> initalized = true;<a name="line.157"></a>
-<span class="sourceLineNo">158</span> }<a name="line.158"></a>
-<span class="sourceLineNo">159</span><a name="line.159"></a>
-<span class="sourceLineNo">160</span> private void usage() {<a name="line.160"></a>
-<span class="sourceLineNo">161</span> System.err.println("usage: " + NAME + " /path/to/hfileoutputformat-output tablename" + "\n -D"<a name="line.161"></a>
-<span class="sourceLineNo">162</span> + CREATE_TABLE_CONF_KEY + "=no - can be used to avoid creation of table by this tool\n"<a name="line.162"></a>
-<span class="sourceLineNo">163</span> + " Note: if you set this to 'no', then the target table must already exist in HBase\n"<a name="line.163"></a>
-<span class="sourceLineNo">164</span> + "\n");<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> private interface BulkHFileVisitor<TFamily> {<a name="line.167"></a>
-<span class="sourceLineNo">168</span> TFamily bulkFamily(final byte[] familyName)<a name="line.168"></a>
-<span class="sourceLineNo">169</span> throws IOException;<a name="line.169"></a>
-<span class="sourceLineNo">170</span> void bulkHFile(final TFamily family, final FileStatus hfileStatus)<a name="line.170"></a>
-<span class="sourceLineNo">171</span> throws IOException;<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> * Iterate over the bulkDir hfiles.<a name="line.175"></a>
-<span class="sourceLineNo">176</span> * Skip reference, HFileLink, files starting with "_" and non-valid hfiles.<a name="line.176"></a>
-<span class="sourceLineNo">177</span> */<a name="line.177"></a>
-<span class="sourceLineNo">178</span> private static <TFamily> void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.178"></a>
-<span class="sourceLineNo">179</span> final BulkHFileVisitor<TFamily> visitor) throws IOException {<a name="line.179"></a>
-<span class="sourceLineNo">180</span> visitBulkHFiles(fs, bulkDir, visitor, true);<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> /**<a name="line.183"></a>
-<span class="sourceLineNo">184</span> * Iterate over the bulkDir hfiles.<a name="line.184"></a>
-<span class="sourceLineNo">185</span> * Skip reference, HFileLink, files starting with "_".<a name="line.185"></a>
-<span class="sourceLineNo">186</span> * Check and skip non-valid hfiles by default, or skip this validation by setting<a name="line.186"></a>
-<span class="sourceLineNo">187</span> * 'hbase.loadincremental.validate.hfile' to false.<a name="line.187"></a>
-<span class="sourceLineNo">188</span> */<a name="line.188"></a>
-<span class="sourceLineNo">189</span> private static <TFamily> void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.189"></a>
-<span class="sourceLineNo">190</span> final BulkHFileVisitor<TFamily> visitor, final boolean validateHFile) throws IOException {<a name="line.190"></a>
-<span class="sourceLineNo">191</span> if (!fs.exists(bulkDir)) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> throw new FileNotFoundException("Bulkload dir " + bulkDir + " not found");<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> FileStatus[] familyDirStatuses = fs.listStatus(bulkDir);<a name="line.195"></a>
-<span class="sourceLineNo">196</span> if (familyDirStatuses == null) {<a name="line.196"></a>
-<span class="sourceLineNo">197</span> throw new FileNotFoundException("No families found in " + bulkDir);<a name="line.197"></a>
-<span class="sourceLineNo">198</span> }<a name="line.198"></a>
-<span class="sourceLineNo">199</span><a name="line.199"></a>
-<span class="sourceLineNo">200</span> for (FileStatus familyStat : familyDirStatuses) {<a name="line.200"></a>
-<span class="sourceLineNo">201</span> if (!familyStat.isDirectory()) {<a name="line.201"></a>
-<span class="sourceLineNo">202</span> LOG.warn("Skipping non-directory " + familyStat.getPath());<a name="line.202"></a>
-<span class="sourceLineNo">203</span> continue;<a name="line.203"></a>
-<span class="sourceLineNo">204</span> }<a name="line.204"></a>
-<span class="sourceLineNo">205</span> Path familyDir = familyStat.getPath();<a name="line.205"></a>
-<span class="sourceLineNo">206</span> byte[] familyName = familyDir.getName().getBytes();<a name="line.206"></a>
-<span class="sourceLineNo">207</span> // Skip invalid family<a name="line.207"></a>
-<span class="sourceLineNo">208</span> try {<a name="line.208"></a>
-<span class="sourceLineNo">209</span> HColumnDescriptor.isLegalFamilyName(familyName);<a name="line.209"></a>
-<span class="sourceLineNo">210</span> }<a name="line.210"></a>
-<span class="sourceLineNo">211</span> catch (IllegalArgumentException e) {<a name="line.211"></a>
-<span class="sourceLineNo">212</span> LOG.warn("Skipping invalid " + familyStat.getPath());<a name="line.212"></a>
-<span class="sourceLineNo">213</span> continue;<a name="line.213"></a>
-<span class="sourceLineNo">214</span> }<a name="line.214"></a>
-<span class="sourceLineNo">215</span> TFamily family = visitor.bulkFamily(familyName);<a name="line.215"></a>
-<span class="sourceLineNo">216</span><a name="line.216"></a>
-<span class="sourceLineNo">217</span> FileStatus[] hfileStatuses = fs.listStatus(familyDir);<a name="line.217"></a>
-<span class="sourceLineNo">218</span> for (FileStatus hfileStatus : hfileStatuses) {<a name="line.218"></a>
-<span class="sourceLineNo">219</span> if (!fs.isFile(hfileStatus.getPath())) {<a name="line.219"></a>
-<span class="sourceLineNo">220</span> LOG.warn("Skipping non-file " + hfileStatus);<a name="line.220"></a>
-<span class="sourceLineNo">221</span> continue;<a name="line.221"></a>
-<span class="sourceLineNo">222</span> }<a name="line.222"></a>
-<span class="sourceLineNo">223</span><a name="line.223"></a>
-<span class="sourceLineNo">224</span> Path hfile = hfileStatus.getPath();<a name="line.224"></a>
-<span class="sourceLineNo">225</span> // Skip "_", reference, HFileLink<a name="line.225"></a>
-<span class="sourceLineNo">226</span> String fileName = hfile.getName();<a name="line.226"></a>
-<span class="sourceLineNo">227</span> if (fileName.startsWith("_")) {<a name="line.227"></a>
-<span class="sourceLineNo">228</span> continue;<a name="line.228"></a>
-<span class="sourceLineNo">229</span> }<a name="line.229"></a>
-<span class="sourceLineNo">230</span> if (StoreFileInfo.isReference(fileName)) {<a name="line.230"></a>
-<span class="sourceLineNo">231</span> LOG.warn("Skipping reference " + fileName);<a name="line.231"></a>
-<span class="sourceLineNo">232</span> continue;<a name="line.232"></a>
-<span class="sourceLineNo">233</span> }<a name="line.233"></a>
-<span class="sourceLineNo">234</span> if (HFileLink.isHFileLink(fileName)) {<a name="line.234"></a>
-<span class="sourceLineNo">235</span> LOG.warn("Skipping HFileLink " + fileName);<a name="line.235"></a>
-<span class="sourceLineNo">236</span> continue;<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> // Validate HFile Format if needed<a name="line.239"></a>
-<span class="sourceLineNo">240</span> if (validateHFile) {<a name="line.240"></a>
-<span class="sourceLineNo">241</span> try {<a name="line.241"></a>
-<span class="sourceLineNo">242</span> if (!HFile.isHFileFormat(fs, hfile)) {<a name="line.242"></a>
-<span class="sourceLineNo">243</span> LOG.warn("the file " + hfile + " doesn't seems to be an hfile. skipping");<a name="line.243"></a>
-<span class="sourceLineNo">244</span> continue;<a name="line.244"></a>
-<span class="sourceLineNo">245</span> }<a name="line.245"></a>
-<span class="sourceLineNo">246</span> } catch (FileNotFoundException e) {<a name="line.246"></a>
-<span class="sourceLineNo">247</span> LOG.warn("the file " + hfile + " was removed");<a name="line.247"></a>
-<span class="sourceLineNo">248</span> continue;<a name="line.248"></a>
-<span class="sourceLineNo">249</span> }<a name="line.249"></a>
-<span class="sourceLineNo">250</span> }<a name="line.250"></a>
-<span class="sourceLineNo">251</span><a name="line.251"></a>
-<span class="sourceLineNo">252</span> visitor.bulkHFile(family, hfileStatus);<a name="line.252"></a>
-<span class="sourceLineNo">253</span> }<a name="line.253"></a>
-<span class="sourceLineNo">254</span> }<a name="line.254"></a>
-<span class="sourceLineNo">255</span> }<a name="line.255"></a>
-<span class="sourceLineNo">256</span><a name="line.256"></a>
-<span class="sourceLineNo">257</span> /**<a name="line.257"></a>
-<span class="sourceLineNo">258</span> * Represents an HFile waiting to be loaded. An queue is used<a name="line.258"></a>
-<span class="sourceLineNo">259</span> * in this class in order to support the case where a region has<a name="line.259"></a>
-<span class="sourceLineNo">260</span> * split during the process of the load. When this happens,<a name="line.260"></a>
-<span class="sourceLineNo">261</span> * the HFile is split into two physical parts across the new<a name="line.261"></a>
-<span class="sourceLineNo">262</span> * region boundary, and each part is added back into the queue.<a name="line.262"></a>
-<span class="sourceLineNo">263</span> * The import process finishes when the queue is empty.<a name="line.263"></a>
-<span class="sourceLineNo">264</span> */<a name="line.264"></a>
-<span class="sourceLineNo">265</span> public static class LoadQueueItem {<a name="line.265"></a>
-<span class="sourceLineNo">266</span> final byte[] family;<a name="line.266"></a>
-<span class="sourceLineNo">267</span> final Path hfilePath;<a name="line.267"></a>
-<span class="sourceLineNo">268</span><a name="line.268"></a>
-<span class="sourceLineNo">269</span> public LoadQueueItem(byte[] family, Path hfilePath) {<a name="line.269"></a>
-<span class="sourceLineNo">270</span> this.family = family;<a name="line.270"></a>
-<span class="sourceLineNo">271</span> this.hfilePath = hfilePath;<a name="line.271"></a>
-<span class="sourceLineNo">272</span> }<a name="line.272"></a>
-<span class="sourceLineNo">273</span><a name="line.273"></a>
-<span class="sourceLineNo">274</span> @Override<a name="line.274"></a>
-<span class="sourceLineNo">275</span> public String toString() {<a name="line.275"></a>
-<span class="sourceLineNo">276</span> return "family:"+ Bytes.toString(family) + " path:" + hfilePath.toString();<a name="line.276"></a>
-<span class="sourceLineNo">277</span> }<a name="line.277"></a>
-<span class="sourceLineNo">278</span> }<a name="line.278"></a>
-<span class="sourceLineNo">279</span><a name="line.279"></a>
-<span class="sourceLineNo">280</span> /**<a name="line.280"></a>
-<span class="sourceLineNo">281</span> * Walk the given directory for all HFiles, and return a Queue<a name="line.281"></a>
-<span class="sourceLineNo">282</span> * containing all such files.<a name="line.282"></a>
-<span class="sourceLineNo">283</span> */<a name="line.283"></a>
-<span class="sourceLineNo">284</span> private void discoverLoadQueue(final Deque<LoadQueueItem> ret, final Path hfofDir,<a name="line.284"></a>
-<span class="sourceLineNo">285</span> final boolean validateHFile) throws IOException {<a name="line.285"></a>
-<span class="sourceLineNo">286</span> fs = hfofDir.getFileSystem(getConf());<a name="line.286"></a>
-<span class="sourceLineNo">287</span> visitBulkHFiles(fs, hfofDir, new BulkHFileVisitor<byte[]>() {<a name="line.287"></a>
-<span class="sourceLineNo">288</span> @Override<a name="line.288"></a>
-<span class="sourceLineNo">289</span> public byte[] bulkFamily(final byte[] familyName) {<a name="line.289"></a>
-<span class="sourceLineNo">290</span> return familyName;<a name="line.290"></a>
-<span class="sourceLineNo">291</span> }<a name="line.291"></a>
-<span class="sourceLineNo">292</span> @Override<a name="line.292"></a>
-<span class="sourceLineNo">293</span> public void bulkHFile(final byte[] family, final FileStatus hfile) throws IOException {<a name="line.293"></a>
-<span class="sourceLineNo">294</span> long length = hfile.getLen();<a name="line.294"></a>
-<span class="sourceLineNo">295</span> if (length > getConf().getLong(HConstants.HREGION_MAX_FILESIZE,<a name="line.295"></a>
-<span class="sourceLineNo">296</span> HConstants.DEFAULT_MAX_FILE_SIZE)) {<a name="line.296"></a>
-<span class="sourceLineNo">297</span> LOG.warn("Trying to bulk load hfile " + hfile.getPath() + " with size: " +<a name="line.297"></a>
-<span class="sourceLineNo">298</span> length + " bytes can be problematic as it may lead to oversplitting.");<a name="line.298"></a>
-<span class="sourceLineNo">299</span> }<a name="line.299"></a>
-<span class="sourceLineNo">300</span> ret.add(new LoadQueueItem(family, hfile.getPath()));<a name="line.300"></a>
-<span class="sourceLineNo">301</span> }<a name="line.301"></a>
-<span class="sourceLineNo">302</span> }, validateHFile);<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> * Perform a bulk load of the given directory into the given<a name="line.306"></a>
-<span class="sourceLineNo">307</span> * pre-existing table. This method is not threadsafe.<a name="line.307"></a>
-<span class="sourceLineNo">308</span> *<a name="line.308"></a>
-<span class="sourceLineNo">309</span> * @param hfofDir the directory that was provided as the output path<a name="line.309"></a>
-<span class="sourceLineNo">310</span> * of a job using HFileOutputFormat<a name="line.310"></a>
-<span class="sourceLineNo">311</span> * @param table the table to load into<a name="line.311"></a>
-<span class="sourceLineNo">312</span> * @throws TableNotFoundException if table does not yet exist<a name="line.312"></a>
-<span class="sourceLineNo">313</span> */<a name="line.313"></a>
-<span class="sourceLineNo">314</span> public void doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.314"></a>
-<span class="sourceLineNo">315</span> RegionLocator regionLocator) throws TableNotFoundException, IOException {<a name="line.315"></a>
-<span class="sourceLineNo">316</span><a name="line.316"></a>
-<span class="sourceLineNo">317</span> if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.317"></a>
-<span class="sourceLineNo">318</span> throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.318"></a>
-<span class="sourceLineNo">319</span> }<a name="line.319"></a>
-<span class="sourceLineNo">320</span><a name="line.320"></a>
-<span class="sourceLineNo">321</span> ExecutorService pool = createExecutorService();<a name="line.321"></a>
-<span class="sourceLineNo">322</span><a name="line.322"></a>
-<span class="sourceLineNo">323</span> // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.323"></a>
-<span class="sourceLineNo">324</span> // happen in this thread<a name="line.324"></a>
-<span class="sourceLineNo">325</span> Deque<LoadQueueItem> queue = new LinkedList<>();<a name="line.325"></a>
-<span class="sourceLineNo">326</span> SecureBulkLoadClient secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.326"></a>
-<span class="sourceLineNo">327</span><a name="line.327"></a>
-<span class="sourceLineNo">328</span> try {<a name="line.328"></a>
-<span class="sourceLineNo">329</span> /*<a name="line.329"></a>
-<span class="sourceLineNo">330</span> * Checking hfile format is a time-consuming operation, we should have an option to skip<a name="line.330"></a>
-<span class="sourceLineNo">331</span> * this step when bulkloading millions of HFiles. See HBASE-13985.<a name="line.331"></a>
-<span class="sourceLineNo">332</span> */<a name="line.332"></a>
-<span class="sourceLineNo">333</span> boolean validateHFile = getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a name="line.333"></a>
-<span class="sourceLineNo">334</span> if(!validateHFile) {<a name="line.334"></a>
-<span class="sourceLineNo">335</span> LOG.warn("You are skipping HFiles validation, it might cause some data loss if files " +<a name="line.335"></a>
-<span class="sourceLineNo">336</span> "are not correct. If you fail to read data from your table after using this " +<a name="line.336"></a>
-<span class="sourceLineNo">337</span> "option, consider removing the files and bulkload again without this option. " +<a name="line.337"></a>
-<span class="sourceLineNo">338</span> "See HBASE-13985");<a name="line.338"></a>
-<span class="sourceLineNo">339</span> }<a name="line.339"></a>
-<span class="sourceLineNo">340</span> prepareHFileQueue(hfofDir, table, queue, validateHFile);<a name="line.340"></a>
-<span class="sourceLineNo">341</span><a name="line.341"></a>
-<span class="sourceLineNo">342</span> int count = 0;<a name="line.342"></a>
-<span class="sourceLineNo">343</span><a name="line.343"></a>
-<span class="sourceLineNo">344</span> if (queue.isEmpty()) {<a name="line.344"></a>
-<span class="sourceLineNo">345</span> LOG.warn("Bulk load operation did not find any files to load in " +<a name="line.345"></a>
-<span class="sourceLineNo">346</span> "directory " + hfofDir.toUri() + ". Does it contain files in " +<a name="line.346"></a>
-<span class="sourceLineNo">347</span> "subdirectories that correspond to column family names?");<a name="line.347"></a>
-<span class="sourceLineNo">348</span> return;<a name="line.348"></a>
-<span class="sourceLineNo">349</span> }<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> if(isSecureBulkLoadEndpointAvailable()) {<a name="line.351"></a>
-<span class="sourceLineNo">352</span> LOG.warn("SecureBulkLoadEndpoint is deprecated. It will be removed in future releases.");<a name="line.352"></a>
-<span class="sourceLineNo">353</span> LOG.warn("Secure bulk load has been integrated into HBase core.");<a name="line.353"></a>
-<span class="sourceLineNo">354</span> }<a name="line.354"></a>
-<span class="sourceLineNo">355</span><a name="line.355"></a>
-<span class="sourceLineNo">356</span> //If using secure bulk load, get source delegation token, and<a name="line.356"></a>
-<span class="sourceLineNo">357</span> //prepare staging directory and token<a name="line.357"></a>
-<span class="sourceLineNo">358</span> // fs is the source filesystem<a name="line.358"></a>
-<span class="sourceLineNo">359</span> fsDelegationToken.acquireDelegationToken(fs);<a name="line.359"></a>
-<span class="sourceLineNo">360</span> bulkToken = secureClient.prepareBulkLoad(admin.getConnection());<a name="line.360"></a>
-<span class="sourceLineNo">361</span><a name="line.361"></a>
-<span class="sourceLineNo">362</span> // Assumes that region splits can happen while this occurs.<a name="line.362"></a>
-<span class="sourceLineNo">363</span> while (!queue.isEmpty()) {<a name="line.363"></a>
-<span class="sourceLineNo">364</span> // need to reload split keys each iteration.<a name="line.364"></a>
-<span class="sourceLineNo">365</span> final Pair<byte[][], byte[][]> startEndKeys = regionLocator.getStartEndKeys();<a name="line.365"></a>
-<span class="sourceLineNo">366</span> if (count != 0) {<a name="line.366"></a>
-<span class="sourceLineNo">367</span> LOG.info("Split occured while grouping HFiles, retry attempt " +<a name="line.367"></a>
-<span class="sourceLineNo">368</span> + count + " with " + queue.size() + " files remaining to group or split");<a name="line.368"></a>
-<span class="sourceLineNo">369</span> }<a name="line.369"></a>
-<span class="sourceLineNo">370</span><a name="line.370"></a>
-<span class="sourceLineNo">371</span> int maxRetries = getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.371"></a>
-<span class="sourceLineNo">372</span> maxRetries = Math.max(maxRetries, startEndKeys.getFirst().length + 1);<a name="line.372"></a>
-<span class="sourceLineNo">373</span> if (maxRetries != 0 && count >= maxRetries) {<a name="line.373"></a>
-<span class="sourceLineNo">374</span> throw new IOException("Retry attempted " + count +<a name="line.374"></a>
-<span class="sourceLineNo">375</span> " times without completing, bailing out");<a name="line.375"></a>
-<span class="sourceLineNo">376</span> }<a name="line.376"></a>
-<span class="sourceLineNo">377</span> count++;<a name="line.377"></a>
-<span class="sourceLineNo">378</span><a name="line.378"></a>
-<span class="sourceLineNo">379</span> // Using ByteBuffer for byte[] equality semantics<a name="line.379"></a>
-<span class="sourceLineNo">380</span> Multimap<ByteBuffer, LoadQueueItem> regionGroups = groupOrSplitPhase(table,<a name="line.380"></a>
-<span class="sourceLineNo">381</span> pool, queue, startEndKeys);<a name="line.381"></a>
-<span class="sourceLineNo">382</span><a name="line.382"></a>
-<span class="sourceLineNo">383</span> if (!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.383"></a>
-<span class="sourceLineNo">384</span> // Error is logged inside checkHFilesCountPerRegionPerFamily.<a name="line.384"></a>
-<span class="sourceLineNo">385</span> throw new IOException("Trying to load more than " + maxFilesPerRegionPerFamily<a name="line.385"></a>
-<span class="sourceLineNo">386</span> + " hfiles to one family of one region");<a name="line.386"></a>
-<span class="sourceLineNo">387</span> }<a name="line.387"></a>
-<span class="sourceLineNo">388</span><a name="line.388"></a>
-<span class="sourceLineNo">389</span> bulkLoadPhase(table, admin.getConnection(), pool, queue, regionGroups);<a name="line.389"></a>
-<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span> // NOTE: The next iteration's split / group could happen in parallel to<a name="line.391"></a>
-<span class="sourceLineNo">392</span> // atomic bulkloads assuming that there are splits and no merges, and<a name="line.392"></a>
-<span class="sourceLineNo">393</span> // that we can atomically pull out the groups we want to retry.<a name="line.393"></a>
-<span class="sourceLineNo">394</span> }<a name="line.394"></a>
-<span class="sourceLineNo">395</span><a name="line.395"></a>
-<span class="sourceLineNo">396</span> } finally {<a name="line.396"></a>
-<span class="sourceLineNo">397</span> fsDelegationToken.releaseDelegationToken();<a name="line.397"></a>
-<span class="sourceLineNo">398</span> if(bulkToken != null) {<a name="line.398"></a>
-<span class="sourceLineNo">399</span> secureClient.cleanupBulkLoad(admin.getConnection(), bulkToken);<a name="line.399"></a>
-<span class="sourceLineNo">400</span> }<a name="line.400"></a>
-<span class="sourceLineNo">401</span> pool.shutdown();<a name="line.401"></a>
-<span class="sourceLineNo">402</span> if (queue != null && !queue.isEmpty()) {<a name="line.402"></a>
-<span class="sourceLineNo">403</span> StringBuilder err = new StringBuilder();<a name="line.403"></a>
-<span class="sourceLineNo">404</span> err.append("-------------------------------------------------\n");<a name="line.404"></a>
-<span class="sourceLineNo">405</span> err.append("Bulk load aborted with some files not yet loaded:\n");<a name="line.405"></a>
-<span class="sourceLineNo">406</span> err.append("-------------------------------------------------\n");<a name="line.406"></a>
-<span class="sourceLineNo">407</span> for (LoadQueueItem q : queue) {<a name="line.407"></a>
-<span class="sourceLineNo">408</span> err.append(" ").append(q.hfilePath).append('\n');<a name="line.408"></a>
+<span class="sourceLineNo">136</span> private LoadIncrementalHFiles() {}<a name="line.136"></a>
+<span class="sourceLineNo">137</span><a name="line.137"></a>
+<span class="sourceLineNo">138</span> public LoadIncrementalHFiles(Configuration conf) throws Exception {<a name="line.138"></a>
+<span class="sourceLineNo">139</span> super(conf);<a name="line.139"></a>
+<span class="sourceLineNo">140</span> this.rpcControllerFactory = new RpcControllerFactory(conf);<a name="line.140"></a>
+<span class="sourceLineNo">141</span> initialize();<a name="line.141"></a>
+<span class="sourceLineNo">142</span> }<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> private void initialize() throws Exception {<a name="line.144"></a>
+<span class="sourceLineNo">145</span> if (initalized) {<a name="line.145"></a>
+<span class="sourceLineNo">146</span> return;<a name="line.146"></a>
+<span class="sourceLineNo">147</span> }<a name="line.147"></a>
+<span class="sourceLineNo">148</span> // make a copy, just to be sure we're not overriding someone else's config<a name="line.148"></a>
+<span class="sourceLineNo">149</span> setConf(HBaseConfiguration.create(getConf()));<a name="line.149"></a>
+<span class="sourceLineNo">150</span> Configuration conf = getConf();<a name="line.150"></a>
+<span class="sourceLineNo">151</span> // disable blockcache for tool invocation, see HBASE-10500<a name="line.151"></a>
+<span class="sourceLineNo">152</span> conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.152"></a>
+<span class="sourceLineNo">153</span> this.userProvider = UserProvider.instantiate(conf);<a name="line.153"></a>
+<span class="sourceLineNo">154</span> this.fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.154"></a>
+<span class="sourceLineNo">155</span> assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.155"></a>
+<span class="sourceLineNo">156</span> maxFilesPerRegionPerFamily = conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.156"></a>
+<span class="sourceLineNo">157</span> nrThreads = conf.getInt("hbase.loadincremental.threads.max",<a name="line.157"></a>
+<span class="sourceLineNo">158</span> Runtime.getRuntime().availableProcessors());<a name="line.158"></a>
+<span class="sourceLineNo">159</span> initalized = true;<a name="line.159"></a>
+<span class="sourceLineNo">160</span> }<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> private void usage() {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> System.err.println("usage: " + NAME + " /path/to/hfileoutputformat-output tablename" + "\n -D"<a name="line.163"></a>
+<span class="sourceLineNo">164</span> + CREATE_TABLE_CONF_KEY + "=no - can be used to avoid creation of table by this tool\n"<a name="line.164"></a>
+<span class="sourceLineNo">165</span> + " Note: if you set this to 'no', then the target table must already exist in HBase\n -D"<a name="line.165"></a>
+<span class="sourceLineNo">166</span> + SILENCE_CONF_KEY + "=yes - can be used to ignore unmatched column families\n"<a name="line.166"></a>
+<span class="sourceLineNo">167</span> + "\n");<a name="line.167"></a>
+<span class="sourceLineNo">168</span> }<a name="line.168"></a>
+<span class="sourceLineNo">169</span><a name="line.169"></a>
+<span class="sourceLineNo">170</span> private interface BulkHFileVisitor<TFamily> {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> TFamily bulkFamily(final byte[] familyName)<a name="line.171"></a>
+<span class="sourceLineNo">172</span> throws IOException;<a name="line.172"></a>
+<span class="sourceLineNo">173</span> void bulkHFile(final TFamily family, final FileStatus hfileStatus)<a name="line.173"></a>
+<span class="sourceLineNo">174</span> throws IOException;<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> /**<a name="line.177"></a>
+<span class="sourceLineNo">178</span> * Iterate over the bulkDir hfiles.<a name="line.178"></a>
+<span class="sourceLineNo">179</span> * Skip reference, HFileLink, files starting with "_" and non-valid hfiles.<a name="line.179"></a>
+<span class="sourceLineNo">180</span> */<a name="line.180"></a>
+<span class="sourceLineNo">181</span> private static <TFamily> void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.181"></a>
+<span class="sourceLineNo">182</span> final BulkHFileVisitor<TFamily> visitor) throws IOException {<a name="line.182"></a>
+<span class="sourceLineNo">183</span> visitBulkHFiles(fs, bulkDir, visitor, true);<a name="line.183"></a>
+<span class="sourceLineNo">184</span> }<a name="line.184"></a>
+<span class="sourceLineNo">185</span><a name="line.185"></a>
+<span class="sourceLineNo">186</span> /**<a name="line.186"></a>
+<span class="sourceLineNo">187</span> * Iterate over the bulkDir hfiles.<a name="line.187"></a>
+<span class="sourceLineNo">188</span> * Skip reference, HFileLink, files starting with "_".<a name="line.188"></a>
+<span class="sourceLineNo">189</span> * Check and skip non-valid hfiles by default, or skip this validation by setting<a name="line.189"></a>
+<span class="sourceLineNo">190</span> * 'hbase.loadincremental.validate.hfile' to false.<a name="line.190"></a>
+<span class="sourceLineNo">191</span> */<a name="line.191"></a>
+<span class="sourceLineNo">192</span> private static <TFamily> void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.192"></a>
+<span class="sourceLineNo">193</span> final BulkHFileVisitor<TFamily> visitor, final boolean validateHFile) throws IOException {<a name="line.193"></a>
+<span class="sourceLineNo">194</span> if (!fs.exists(bulkDir)) {<a name="line.194"></a>
+<span class="sourceLineNo">195</span> throw new FileNotFoundException("Bulkload dir " + bulkDir + " not found");<a name="line.195"></a>
+<span class="sourceLineNo">196</span> }<a name="line.196"></a>
+<span class="sourceLineNo">197</span><a name="line.197"></a>
+<span class="sourceLineNo">198</span> FileStatus[] familyDirStatuses = fs.listStatus(bulkDir);<a name="line.198"></a>
+<span class="sourceLineNo">199</span> if (familyDirStatuses == null) {<a name="line.199"></a>
+<span class="sourceLineNo">200</span> throw new FileNotFoundException("No families found in " + bulkDir);<a name="line.200"></a>
+<span class="sourceLineNo">201</span> }<a name="line.201"></a>
+<span class="sourceLineNo">202</span><a name="line.202"></a>
+<span class="sourceLineNo">203</span> for (FileStatus familyStat : familyDirStatuses) {<a name="line.203"></a>
+<span class="sourceLineNo">204</span> if (!familyStat.isDirectory()) {<a name="line.204"></a>
+<span class="sourceLineNo">205</span> LOG.warn("Skipping non-directory " + familyStat.getPath());<a name="line.205"></a>
+<span class="sourceLineNo">206</span> continue;<a name="line.206"></a>
+<span class="sourceLineNo">207</span> }<a name="line.207"></a>
+<span class="sourceLineNo">208</span> Path familyDir = familyStat.getPath();<a name="line.208"></a>
+<span class="sourceLineNo">209</span> byte[] familyName = familyDir.getName().getBytes();<a name="line.209"></a>
+<span class="sourceLineNo">210</span> // Skip invalid family<a name="line.210"></a>
+<span class="sourceLineNo">211</span> try {<a name="line.211"></a>
+<span class="sourceLineNo">212</span> HColumnDescriptor.isLegalFamilyName(familyName);<a name="line.212"></a>
+<span class="sourceLineNo">213</span> }<a name="line.213"></a>
+<span class="sourceLineNo">214</span> catch (IllegalArgumentException e) {<a name="line.214"></a>
+<span class="sourceLineNo">215</span> LOG.warn("Skipping invalid " + familyStat.getPath());<a name="line.215"></a>
+<span class="sourceLineNo">216</span> continue;<a name="line.216"></a>
+<span class="sourceLineNo">217</span> }<a name="line.217"></a>
+<span class="sourceLineNo">218</span> TFamily family = visitor.bulkFamily(familyName);<a name="line.218"></a>
+<span class="sourceLineNo">219</span><a name="line.219"></a>
+<span class="sourceLineNo">220</span> FileStatus[] hfileStatuses = fs.listStatus(familyDir);<a name="line.220"></a>
+<span class="sourceLineNo">221</span> for (FileStatus hfileStatus : hfileStatuses) {<a name="line.221"></a>
+<span class="sourceLineNo">222</span> if (!fs.isFile(hfileStatus.getPath())) {<a name="line.222"></a>
+<span class="sourceLineNo">223</span> LOG.warn("Skipping non-file " + hfileStatus);<a name="line.223"></a>
+<span class="sourceLineNo">224</span> continue;<a name="line.224"></a>
+<span class="sourceLineNo">225</span> }<a name="line.225"></a>
+<span class="sourceLineNo">226</span><a name="line.226"></a>
+<span class="sourceLineNo">227</span> Path hfile = hfileStatus.getPath();<a name="line.227"></a>
+<span class="sourceLineNo">228</span> // Skip "_", reference, HFileLink<a name="line.228"></a>
+<span class="sourceLineNo">229</span> String fileName = hfile.getName();<a name="line.229"></a>
+<span class="sourceLineNo">230</span> if (fileName.startsWith("_")) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span> continue;<a name="line.231"></a>
+<span class="sourceLineNo">232</span> }<a name="line.232"></a>
+<span class="sourceLineNo">233</span> if (StoreFileInfo.isReference(fileName)) {<a name="line.233"></a>
+<span class="sourceLineNo">234</span> LOG.warn("Skipping reference " + fileName);<a name="line.234"></a>
+<span class="sourceLineNo">235</span> continue;<a name="line.235"></a>
+<span class="sourceLineNo">236</span> }<a name="line.236"></a>
+<span class="sourceLineNo">237</span> if (HFileLink.isHFileLink(fileName)) {<a name="line.237"></a>
+<span class="sourceLineNo">238</span> LOG.warn("Skipping HFileLink " + fileName);<a name="line.238"></a>
+<span class="sourceLineNo">239</span> continue;<a name="line.239"></a>
+<span class="sourceLineNo">240</span> }<a name="line.240"></a>
+<span class="sourceLineNo">241</span><a name="line.241"></a>
+<span class="sourceLineNo">242</span> // Validate HFile Format if needed<a name="line.242"></a>
+<span class="sourceLineNo">243</span> if (validateHFile) {<a name="line.243"></a>
+<span class="sourceLineNo">244</span> try {<a name="line.244"></a>
+<span class="sourceLineNo">245</span> if (!HFile.isHFileFormat(fs, hfile)) {<a name="line.245"></a>
+<span class="sourceLineNo">246</span> LOG.warn("the file " + hfile + " doesn't seems to be an hfile. skipping");<a name="line.246"></a>
+<span class="sourceLineNo">247</span> continue;<a name="line.247"></a>
+<span class="sourceLineNo">248</span> }<a name="line.248"></a>
+<span class="sourceLineNo">249</span> } catch (FileNotFoundException e) {<a name="line.249"></a>
+<span class="sourceLineNo">250</span> LOG.warn("the file " + hfile + " was removed");<a name="line.250"></a>
+<span class="sourceLineNo">251</span> continue;<a name="line.251"></a>
+<span class="sourceLineNo">252</span> }<a name="line.252"></a>
+<span class="sourceLineNo">253</span> }<a name="line.253"></a>
+<span class="sourceLineNo">254</span><a name="line.254"></a>
+<span class="sourceLineNo">255</span> visitor.bulkHFile(family, hfileStatus);<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> }<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> * Represents an HFile waiting to be loaded. An queue is used<a name="line.261"></a>
+<span class="sourceLineNo">262</span> * in this class in order to support the case where a region has<a name="line.262"></a>
+<span class="sourceLineNo">263</span> * split during the process of the load. When this happens,<a name="line.263"></a>
+<span class="sourceLineNo">264</span> * the HFile is split into two physical parts across the new<a name="line.264"></a>
+<span class="sourceLineNo">265</span> * region boundary, and each part is added back into the queue.<a name="line.265"></a>
+<span class="sourceLineNo">266</span> * The import process finishes when the queue is empty.<a name="line.266"></a>
+<span class="sourceLineNo">267</span> */<a name="line.267"></a>
+<span class="sourceLineNo">268</span> public static class LoadQueueItem {<a name="line.268"></a>
+<span class="sourceLineNo">269</span> final byte[] family;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> final Path hfilePath;<a name="line.270"></a>
+<span class="sourceLineNo">271</span><a name="line.271"></a>
+<span class="sourceLineNo">272</span> public LoadQueueItem(byte[] family, Path hfilePath) {<a name="line.272"></a>
+<span class="sourceLineNo">273</span> this.family = family;<a name="line.273"></a>
+<span class="sourceLineNo">274</span> this.hfilePath = hfilePath;<a name="line.274"></a>
+<span class="sourceLineNo">275</span> }<a name="line.275"></a>
+<span class="sourceLineNo">276</span><a name="line.276"></a>
+<span class="sourceLineNo">277</span> @Override<a name="line.277"></a>
+<span class="sourceLineNo">278</span> public String toString() {<a name="line.278"></a>
+<span class="sourceLineNo">279</span> return "family:"+ Bytes.toString(family) + " path:" + hfilePath.toString();<a name="line.279"></a>
+<span class="sourceLineNo">280</span> }<a name="line.280"></a>
+<span class="sourceLineNo">281</span> }<a name="line.281"></a>
+<span class="sourceLineNo">282</span><a name="line.282"></a>
+<span class="sourceLineNo">283</span> /**<a name="line.283"></a>
+<span class="sourceLineNo">284</span> * Walk the given directory for all HFiles, and return a Queue<a name="line.284"></a>
+<span class="sourceLineNo">285</span> * containing all such files.<a name="line.285"></a>
+<span class="sourceLineNo">286</span> */<a name="line.286"></a>
+<span class="sourceLineNo">287</span> private void discoverLoadQueue(final Deque<LoadQueueItem> ret, final Path hfofDir,<a name="line.287"></a>
+<span class="sourceLineNo">288</span> final boolean validateHFile) throws IOException {<a name="line.288"></a>
+<span class="sourceLineNo">289</span> fs = hfofDir.getFileSystem(getConf());<a name="line.289"></a>
+<span class="sourceLineNo">290</span> visitBulkHFiles(fs, hfofDir, new BulkHFileVisitor<byte[]>() {<a name="line.290"></a>
+<span class="sourceLineNo">291</span> @Override<a name="line.291"></a>
+<span class="sourceLineNo">292</span> public byte[] bulkFamily(final byte[] familyName) {<a name="line.292"></a>
+<span class="sourceLineNo">293</span> return familyName;<a name="line.293"></a>
+<span class="sourceLineNo">294</span> }<a name="line.294"></a>
+<span class="sourceLineNo">295</span> @Override<a name="line.295"></a>
+<span class="sourceLineNo">296</span> public void bulkHFile(final byte[] family, final FileStatus hfile) throws IOException {<a name="line.296"></a>
+<span class="sourceLineNo">297</span> long length = hfile.getLen();<a name="line.297"></a>
+<span class="sourceLineNo">298</span> if (length > getConf().getLong(HConstants.HREGION_MAX_FILESIZE,<a name="line.298"></a>
+<span class="sourceLineNo">299</span> HConstants.DEFAULT_MAX_FILE_SIZE)) {<a name="line.299"></a>
+<span class="sourceLineNo">300</span> LOG.warn("Trying to bulk load hfile " + hfile.getPath() + " with size: " +<a name="line.300"></a>
+<span class="sourceLineNo">301</span> length + " bytes can be problematic as it may lead to oversplitting.");<a name="line.301"></a>
+<span class="sourceLineNo">302</span> }<a name="line.302"></a>
+<span class="sourceLineNo">303</span> ret.add(new LoadQueueItem(family, hfile.getPath()));<a name="line.303"></a>
+<span class="sourceLineNo">304</span> }<a name="line.304"></a>
+<span class="sourceLineNo">305</span> }, validateHFile);<a name="line.305"></a>
+<span class="sourceLineNo">306</span> }<a name="line.306"></a>
+<span class="sourceLineNo">307</span><a name="line.307"></a>
+<span class="sourceLineNo">308</span> /**<a name="line.308"></a>
+<span class="sourceLineNo">309</span> * Perform a bulk load of the given directory into the given<a name="line.309"></a>
+<span class="sourceLineNo">310</span> * pre-existing table. This method is not threadsafe.<a name="line.310"></a>
+<span class="sourceLineNo">311</span> *<a name="line.311"></a>
+<span class="sourceLineNo">312</span> * @param hfofDir the directory that was provided as the output path<a name="line.312"></a>
+<span class="sourceLineNo">313</span> * of a job using HFileOutputFormat<a name="line.313"></a>
+<span class="sourceLineNo">314</span> * @param admin the Admin<a name="line.314"></a>
+<span class="sourceLineNo">315</span> * @param table the table to load into<a name="line.315"></a>
+<span class="sourceLineNo">316</span> * @param regionLocator region locator<a name="line.316"></a>
+<span class="sourceLineNo">317</span> * @throws TableNotFoundException if table does not yet exist<a name="line.317"></a>
+<span class="sourceLineNo">318</span> */<a name="line.318"></a>
+<span class="sourceLineNo">319</span> public void doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.319"></a>
+<span class="sourceLineNo">320</span> RegionLocator regionLocator) throws TableNotFoundException, IOException {<a name="line.320"></a>
+<span class="sourceLineNo">321</span> doBulkLoad(hfofDir, admin, table, regionLocator, false);<a name="line.321"></a>
+<span class="sourceLineNo">322</span> }<a name="line.322"></a>
+<span class="sourceLineNo">323</span><a name="line.323"></a>
+<span class="sourceLineNo">324</span> /**<a name="line.324"></a>
+<span class="sourceLineNo">325</span> * Perform a bulk load of the given directory into the given<a name="line.325"></a>
+<span class="sourceLineNo">326</span> * pre-existing table. This method is not threadsafe.<a name="line.326"></a>
+<span class="sourceLineNo">327</span> *<a name="line.327"></a>
+<span class="sourceLineNo">328</span> * @param hfofDir the directory that was provided as the output path<a name="line.328"></a>
+<span class="sourceLineNo">329</span> * of a job using HFileOutputFormat<a name="line.329"></a>
+<span class="sourceLineNo">330</span> * @param admin the Admin<a name="line.330"></a>
+<span class="sourceLineNo">331</span> * @param table the table to load into<a name="line.331"></a>
+<span class="sourceLineNo">332</span> * @param regionLocator region locator<a name="line.332"></a>
+<span class="sourceLineNo">333</span> * @param silence true to ignore unmatched column families<a name="line.333"></a>
+<span class="sourceLineNo">334</span> * @throws TableNotFoundException if table does not yet exist<a name="line.334"></a>
+<span class="sourceLineNo">335</span> */<a name="line.335"></a>
+<span class="sourceLineNo">336</span> public void doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.336"></a>
+<span class="sourceLineNo">337</span> RegionLocator regionLocator, boolean silence) throws TableNotFoundException, IOException {<a name="line.337"></a>
+<span class="sourceLineNo">338</span><a name="line.338"></a>
+<span class="sourceLineNo">339</span> if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.339"></a>
+<span class="sourceLineNo">340</span> throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.340"></a>
+<span class="sourceLineNo">341</span> }<a name="line.341"></a>
+<span class="sourceLineNo">342</span><a name="line.342"></a>
+<span class="sourceLineNo">343</span> ExecutorService pool = createExecutorService();<a name="line.343"></a>
+<span class="sourceLineNo">344</span><a name="line.344"></a>
+<span class="sourceLineNo">345</span> // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.345"></a>
+<span class="sourceLineNo">346</span> // happen in this thread<a name="line.346"></a>
+<span class="sourceLineNo">347</span> Deque<LoadQueueItem> queue = new LinkedList<>();<a name="line.347"></a>
+<span class="sourceLineNo">348</span> SecureBulkLoadClient secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.348"></a>
+<span class="sourceLineNo">349</span><a name="line.349"></a>
+<span class="sourceLineNo">350</span> try {<a name="line.350"></a>
+<span class="sourceLineNo">351</span> /*<a name="line.351"></a>
+<span class="sourceLineNo">352</span> * Checking hfile format is a time-consuming operation, we should have an option to skip<a name="line.352"></a>
+<span class="sourceLineNo">353</span> * this step when bulkloading millions of HFiles. See HBASE-13985.<a name="line.353"></a>
+<span class="sourceLineNo">354</span> */<a name="line.354"></a>
+<span class="sourceLineNo">355</span> boolean validateHFile = getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a name="line.355"></a>
+<span class="sourceLineNo">356</span> if(!validateHFile) {<a name="line.356"></a>
+<span class="sourceLineNo">357</span> LOG.warn("You are skipping HFiles validation, it might cause some data loss if files " +<a name="line.357"></a>
+<span class="sourceLineNo">358</span> "are not correct. If you fail to read data from your table after using this " +<a name="line.358"></a>
+<span class="sourceLineNo">359</span> "option, consider removing the files and bulkload again without this option. " +<a name="line.359"></a>
+<span class="sourceLineNo">360</span> "See HBASE-13985");<a name="line.360"></a>
+<span class="sourceLineNo">361</span> }<a name="line.361"></a>
+<span class="sourceLineNo">362</span> prepareHFileQueue(hfofDir, table, queue, validateHFile, silence);<a name="line.362"></a>
+<span class="sourceLineNo">363</span><a name="line.363"></a>
+<span class="sourceLineNo">364</span> int count = 0;<a name="line.364"></a>
+<span class="sourceLineNo">365</span><a name="line.365"></a>
+<span class="sourceLineNo">366</span> if (queue.isEmpty()) {<a name="line.366"></a>
+<span class="sourceLineNo">367</span> LOG.warn("Bulk load operation did not find any files to load in " +<a name="line.367"></a>
+<span class="sourceLineNo">368</span> "directory " + hfofDir.toUri() + ". Does it contain files in " +<a name="line.368"></a>
+<span class="sourceLineNo">369</span> "subdirectories that correspond to column family names?");<a name="line.369"></a>
+<span class="sourceLineNo">370</span> return;<a name="line.370"></a>
+<span class="sourceLineNo">371</span> }<a name="line.371"></a>
+<span class="sourceLineNo">372</span><a name="line.372"></a>
+<span class="sourceLineNo">373</span> if(isSecureBulkLoadEndpointAvailable()) {<a name="line.373"></a>
+<span class="sourceLineNo">374</span> LOG.warn("SecureBulkLoadEndpoint is deprecated. It will be removed in future releases.");<a name="line.374"></a>
+<span class="sourceLineNo">375</span> LOG.warn("Secure bulk load has been integrated into HBase core.");<a name="line.375"></a>
+<span class="sourceLineNo">376</span> }<a name="line.376"></a>
+<span class="sourceLineNo">377</span><a name="line.377"></a>
+<span class="sourceLineNo">378</span> //If using secure bulk load, get source delegation token, and<a name="line.378"></a>
+<span class="sourceLineNo">379</span> //prepare staging directory and token<a name="line.379"></a>
+<span class="sourceLineNo">380</span> // fs is the source filesystem<a name="line.380"></a>
+<span class="sourceLineNo">381</span> fsDelegationToken.acquireDelegationToken(fs);<a name="line.381"></a>
+<span class="sourceLineNo">382</span> bulkToken = secureClient.prepareBulkLoad(admin.getConnection());<a name="line.382"></a>
+<span class="sourceLineNo">383</span><a name="line.383"></a>
+<span class="sourceLineNo">384</span> // Assumes that region splits can happen while this occurs.<a name="line.384"></a>
+<span class="sourceLineNo">385</span> while (!queue.isEmpty()) {<a name="line.385"></a>
+<span class="sourceLineNo">386</span> // need to reload split keys each iteration.<a name="line.386"></a>
+<span class="sourceLineNo">387</span> final Pair<byte[][], byte[][]> startEndKeys = regionLocator.getStartEndKeys();<a name="line.387"></a>
+<span class="sourceLineNo">388</span> if (count != 0) {<a name="line.388"></a>
+<span class="sourceLineNo">389</span> LOG.info("Split occured while grouping HFiles, retry attempt " +<a name="line.389"></a>
+<span class="sourceLineNo">390</span> + count + " with " + queue.size() + " files remaining to group or split");<a name="line.390"></a>
+<span class="sourceLineNo">391</span> }<a name="line.391"></a>
+<span class="sourceLineNo">392</span><a name="line.392"></a>
+<span class="sourceLineNo">393</span> int maxRetries = getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.393"></a>
+<span class="sourceLineNo">394</span> maxRetries = Math.max(maxRetries, startEndKeys.getFirst().length + 1);<a name="line.394"></a>
+<span class="sourceLineNo">395</span> if (maxRetries != 0 && count >= maxRetries) {<a name="line.395"></a>
+<span class="sourceLineNo">396</span> throw new IOException("Retry attempted " + count +<a name="line.396"></a>
+<span class="sourceLineNo">397</span> " times without completing, bailing out");<a name="line.397"></a>
+<span class="sourceLineNo">398</span> }<a name="line.398"></a>
+<span class="sourceLineNo">399</span> count++;<a name="line.399"></a>
+<span class="sourceLineNo">400</span><a name="line.400"></a>
+<span class="sourceLineNo">401</span> // Using ByteBuffer for byte[] equality semantics<a name="line.401"></a>
+<span class="sourceLineNo">402</span> Multimap<ByteBuffer, LoadQueueItem> regionGroups = groupOrSplitPhase(table,<a name="line.402"></a>
+<span class="sourceLineNo">403</span> pool, queue, startEndKeys);<a name="line.403"></a>
+<span class="sourceLineNo">404</span><a name="line.404"></a>
+<span class="sourceLineNo">405</span> if (!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.405"></a>
+<span class="sourceLineNo">406</span> // Error is logged inside checkHFilesCountPerRegionPerFamily.<a name="line.406"></a>
+<span class="sourceLineNo">407</span> throw new IOException("Trying to load more than " + maxFilesPerRegionPerFamily<a name="line.407"></a>
+<span class="sourceLineNo">408</span> + " hfiles to one family of one region");<a name="line.408"></a>
<span class="sourceLineNo">409</span> }<a name="line.409"></a>
-<span class="sourceLineNo">410</span> LOG.error(err);<a name="line.410"></a>
-<span class="sourceLineNo">411</span> }<a name="line.411"></a>
-<span class="sourceLineNo">412</span> }<a name="line.412"></a>
-<span class="sourceLineNo">413</span><a name="line.413"></a>
-<span class="sourceLineNo">414</span> if (queue != null && !queue.isEmpty()) {<a name="line.414"></a>
-<span class="sourceLineNo">415</span> throw new RuntimeException("Bulk load aborted with some files not yet loaded."<a name="line.415"></a>
-<span class="sourceLineNo">416</span> + "Please check log for more details.");<a name="line.416"></a>
-<span class="sourceLineNo">417</span> }<a name="line.417"></a>
-<span class="sourceLineNo">418</span> }<a name="line.418"></a>
-<span class="sourceLineNo">419</span><a name="line.419"></a>
-<span class="sourceLineNo">420</span> /**<a name="line.420"></a>
-<span class="sourceLineNo">421</span> * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.421"></a>
-<span class="sourceLineNo">422</span> * passed directory and validates whether the prepared queue has all the valid table column<a name="line.422"></a>
-<span class="sourceLineNo">423</span> * families in it.<a name="line.423"></a>
-<span class="sourceLineNo">424</span> * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.424"></a>
-<span class="sourceLineNo">425</span> * @param table table to which hfiles should be loaded<a name="line.425"></a>
-<span class="sourceLineNo">426</span> * @param queue queue which needs to be loaded into the table<a name="line.426"></a>
-<span class="sourceLineNo">427</span> * @param validateHFile if true hfiles will be validated for its format<a name="line.427"></a>
-<span class="sourceLineNo">428</span> * @throws IOException If any I/O or network error occurred<a name="line.428"></a>
-<span class="sourceLineNo">429</span> */<a name="line.429"></a>
-<span class="sourceLineNo">430</span> public void prepareHFileQueue(Path hfilesDir, Table table, Deque<LoadQueueItem> queue,<a name="line.430"></a>
-<span class="sourceLineNo">431</span> boolean validateHFile) throws IOException {<a name="line.431"></a>
-<span class="sourceLineNo">432</span> discoverLoadQueue(queue, hfilesDir, validateHFile);<a name="line.432"></a>
-<span class="sourceLineNo">433</span> validateFamiliesInHFiles(table, queue);<a name="line.433"></a>
-<span class="sourceLineNo">434</span> }<a name="line.434"></a>
+<span class="sourceLineNo">410</span><a name="line.410"></a>
+<span class="sourceLineNo">411</span> bulkLoadPhase(table, admin.getConnection(), pool, queue, regionGroups);<a name="line.411"></a>
+<span class="sourceLineNo">412</span><a name="line.412"></a>
+<span class="sourceLineNo">413</span> // NOTE: The next iteration's split / group could happen in parallel to<a name="line.413"></a>
+<span class="sourceLineNo">414</span> // atomic bulkloads assuming that there are splits and no merges, and<a name="line.414"></a>
+<span class="sourceLineNo">415</span> // that we can atomically pull out the groups we want to retry.<a name="line.415"></a>
+<span class="sourceLineNo">416</span> }<a name="line.416"></a>
+<span class="sourceLineNo">417</span><a name="line.417"></a>
+<span class="sourceLineNo">418</span> } finally {<a name="line.418"></a>
+<span class="sourceLineNo">419</span> fsDelegationToken.releaseDelegationToken();<a name="line.419"></a>
+<span class="sourceLineNo">420</span> if(bulkToken != null) {<a name="line.420"></a>
+<span class="sourceLineNo">421</span> secureClient.cleanupBulkLoad(admin.getConnection(), bulkToken);<a name="line.421"></a>
+<span class="sourceLineNo">422</span> }<a name="line.422"></a>
+<span class="sourceLineNo">423</span> pool.shutdown();<a name="line.423"></a>
+<span class="sourceLineNo">424</span> if (queue != null && !queue.isEmpty()) {<a name="line.424"></a>
+<span class="sourceLineNo">425</span> StringBuilder err = new StringBuilder();<a name="line.425"></a>
+<span class="sourceLineNo">426</span> err.append("-------------------------------------------------\n");<a name="line.426"></a>
+<span class="sourceLineNo">427</span> err.append("Bulk load aborted with some files not yet loaded:\n");<a name="line.427"></a>
+<span class="sourceLineNo">428</span> err.append("-------------------------------------------------\n");<a name="line.428"></a>
+<span class="sourceLineNo">429</span> for (LoadQueueItem q : queue) {<a name="line.429"></a>
+<span class="sourceLineNo">430</span> err.append(" ").append(q.hfilePath).append('\n');<a name="line.430"></a>
+<span class="sourceLineNo">431</span> }<a name="line.431"></a>
+<span class="sourceLineNo">432</span> LOG.error(err);<a name="line.432"></a>
+<span class="sourceLineNo">433</span> }<a name="line.433"></a>
+<span class="sourceLineNo">434</span> }<a name="line.434"></a>
<span class="sourceLineNo">435</span><a name="line.435"></a>
-<span class="sourceLineNo">436</span> // Initialize a thread pool<a name="line.436"></a>
-<span class="sourceLineNo">437</span> private ExecutorService createExecutorService() {<a name="line.437"></a>
-<span class="sourceLineNo">438</span> ThreadFactoryBuilder builder = new ThreadFactoryBuilder();<a name="line.438"></a>
-<span class="sourceLineNo">439</span> builder.setNameFormat("LoadIncrementalHFiles-%1$d");<a name="line.439"></a>
-<span class="sourceLineNo">440</span> ExecutorService pool = new ThreadPoolExecutor(nrThreads, nrThreads, 60, TimeUnit.SECONDS,<a name="line.440"></a>
-<span class="sourceLineNo">441</span> new LinkedBlockingQueue<Runnable>(), builder.build());<a name="line.441"></a>
-<span class="sourceLineNo">442</span> ((ThreadPoolExecutor) pool).allowCoreThreadTimeOut(true);<a name="line.442"></a>
-<span class="sourceLineNo">443</span> return pool;<a name="line.443"></a>
-<span class="sourceLineNo">444</span> }<a name="line.444"></a>
-<span class="sourceLineNo">445</span><a name="line.445"></a>
-<span class="sourceLineNo">446</span> /**<a name="line.446"></a>
-<span class="sourceLineNo">447</span> * Checks whether there is any invalid family name in HFiles to be bulk loaded.<a name="line.447"></a>
-<span class="sourceLineNo">448</span> */<a name="line.448"></a>
-<span class="sourceLineNo">449</span> private void validateFamiliesInHFiles(Table table, Deque<LoadQueueItem> queue)<a name="line.449"></a>
-<span class="sourceLineNo">450</span> throws IOException {<a name="line.450"></a>
-<span class="sourceLineNo">451</span> Collection<HColumnDescriptor> families = table.getTableDescriptor().getFamilies();<a name="line.451"></a>
-<span class="sourceLineNo">452</span> List<String> familyNames = new ArrayList<>(families.size());<a name="line.452"></a>
-<span class="sourceLineNo">453</span> for (HColumnDescriptor family : families) {<a name="line.453"></a>
-<span class="sourceLineNo">454</span> familyNames.add(family.getNameAsString());<a name="line.454"></a>
-<span class="sourceLineNo">455</span> }<a name="line.455"></a>
-<span class="sourceLineNo">456</span> List<String> unmatchedFamilies = new ArrayList<String>();<a name="line.456"></a>
-<span class="sourceLineNo">457</span> Iterator<LoadQueueItem> queueIter = queue.iterator();<a name="line.457"></a>
-<span class="sourceLineNo">458</span> while (queueIter.hasNext()) {<a name="line.458"></a>
-<span class="sourceLineNo">459</span> LoadQueueItem lqi = queueIter.next();<a name="line.459"></a>
-<span class="sourceLineNo">460</span> String familyNameInHFile = Bytes.toString(lqi.family);<a name="line.460"></a>
-<span class="sourceLineNo">461</span> if (!familyNames.contains(familyNameInHFile)) {<a name="line.461"></a>
-<span class="sourceLineNo">462</span> unmatchedFamilies.add(familyNameInHFile);<a name="line.462"></a>
-<span class="sourceLineNo">463</span> }<a name="line.463"></a>
-<span class="sourceLineNo">464</span> }<a name="line.464"></a>
-<span class="sourceLineNo">465</span> if (unmatchedFamilies.size() > 0) {<a name="line.465"></a>
-<span class="sourceLineNo">466</span> String msg =<a name="line.466"></a>
-<span class="sourceLineNo">467</span> "Unmatched family names found: unmatched family names in HFiles to be bulkloaded: "<a name="line.467"></a>
-<span class="sourceLineNo">468</span> + unmatchedFamilies + "; valid family names of table " + table.getName() + " are: "<a name="line.468"></a>
-<span class="sourceLineNo">469</span> + familyNames;<a name="line.469"></a>
-<span class="sourceLineNo">470</span> LOG.error(msg);<a name="line.470"></a>
-<span class="sourceLineNo">471</span> throw new IOException(msg);<a name="line.471"></a>
-<span class="sourceLineNo">472</span> }<a name="line.472"></a>
-<span class="sourceLineNo">473</span> }<a name="line.473"></a>
-<span class="sourceLineNo">474</span><a name="line.474"></a>
-<span class="sourceLineNo">475</span> /**<a name="line.475"></a>
-<span class="sourceLineNo">476</span> * Used by the replication sink to load the hfiles from the source cluster. It does the following,<a name="line.476"></a>
-<span class="sourceLineNo">477</span> * <ol><a name="line.477"></a>
-<span class="sourceLineNo">478</span> * <li>LoadIncrementalHFiles#groupOrSplitPhase(Table, ExecutorService, Deque, Pair)}</li><a name="line.478"></a>
-<span class="sourceLineNo">479</span> * <li>LoadIncrementalHFiles#bulkLoadPhase(Table, Connection, ExecutorService, Deque, Multimap)<a name="line.479"></a>
-<span class="sourceLineNo">480</span> * </li><a name="line.480"></a>
-<span class="sourceLineNo">481</span> * </ol><a name="line.481"></a>
-<span class="sourceLineNo">482</span> * @param table Table to which these hfiles should be loaded to<a name="line.482"></a>
-<span class="sourceLineNo">483</span> * @param conn Connection to use<a name="line.483"></a>
-<span class="sourceLineNo">484</span> * @param queue {@link LoadQueueItem} has hfiles yet to be loaded<a name="line.484"></a>
-<span class="sourceLineNo">485</span> * @param startEndKeys starting and ending row keys of the region<a name="line.485"></a>
+<span class="sourceLineNo">436</span> if (queue != null && !queue.isEmpty()) {<a name="line.436"></a>
+<span class="sourceLineNo">437</span> throw new RuntimeException("Bulk load aborted with some files not yet loaded."<a name="line.437"></a>
+<span class="sourceLineNo">438</span> + "Please check log for more details.");<a name="line.438"></a>
+<span class="sourceLineNo">439</span> }<a name="line.439"></a>
+<span class="sourceLineNo">440</span> }<a name="line.440"></a>
+<span class="sourceLineNo">441</span><a name="line.441"></a>
+<span class="sourceLineNo">442</span> /**<a name="line.442"></a>
+<span class="sourceLineNo">443</span> * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.443"></a>
+<span class="sourceLineNo">444</span> * passed directory and validates whether the prepared queue has all the valid table column<a name="line.444"></a>
+<span class="sourceLineNo">445</span> * families in it.<a name="line.445"></a>
+<span class="sourceLineNo">446</span> * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.446"></a>
+<span class="sourceLineNo">447</span> * @param table table to which hfiles should be loaded<a name="line.447"></a>
+<span class="sourceLineNo">448</span> * @param queue queue which needs to be loaded into the table<a name="line.448"></a>
+<span class="sourceLineNo">449</span> * @param validateHFile if true hfiles will be validated for its format<a name="line.449"></a>
+<span class="sourceLineNo">450</span> * @throws IOException If any I/O or network error occurred<a name="line.450"></a>
+<span class="sourceLineNo">451</span> */<a name="line.451"></a>
+<span class="sourceLineNo">452</span> public void prepareHFileQueue(Path hfilesDir, Table table, Deque<LoadQueueItem> queue,<a name="line.452"></a>
+<span class="sourceLineNo">453</span> boolean validateHFile) throws IOException {<a name="line.453"></a>
+<span class="sourceLineNo">454</span> prepareHFileQueue(hfilesDir, table, queue, validateHFile, false);<a name="line.454"></a>
+<span class="sourceLineNo">455</span> }<a name="line.455"></a>
+<span class="sourceLineNo">456</span><a name="line.456"></a>
+<span class="sourceLineNo">457</span> /**<a name="line.457"></a>
+<span class="sourceLineNo">458</span> * Prepare a collection of {@link LoadQueueItem} from list of source hfiles contained in the<a name="line.458"></a>
+<span class="sourceLineNo">459</span> * passed directory and validates whether the prepared queue has all the valid table column<a name="line.459"></a>
+<span class="sourceLineNo">460</span> * families in it.<a name="line.460"></a>
+<span class="sourceLineNo">461</span> * @param hfilesDir directory containing list of hfiles to be loaded into the table<a name="line.461"></a>
+<span class="sourceLineNo">462</span> * @param table table to which hfiles should be loaded<a name="line.462"></a>
+<span class="sourceLineNo">463</span> * @param queue queue which needs to be loaded into the table<a name="line.463"></a>
+<span class="sourceLineNo">464</span> * @param validateHFile if true hfiles will be validated for its format<a name="line.464"></a>
+<span class="sourceLineNo">465</span> * @param silence true to ignore unmatched column families<a name="line.465"></a>
+<span class="sourceLineNo">466</span> * @throws IOException If any I/O or network error occurred<a name="line.466"></a>
+<span class="sourceLineNo">467</span> */<a name="line.467"></a>
+<span class="sourceLineNo">468</span> public void prepareHFileQueue(Path hfilesDir, Table table, Deque<LoadQueueItem> queue,<a name="line.468"></a>
+<span class="sourceLineNo">469</span> boolean validateHFile, boolean silence) throws IOException {<a name="line.469"></a>
+<span class="sourceLineNo">470</span> discoverLoadQueue(queue, hfilesDir, validateHFile);<a name="line.470"></a>
+<span class="sourceLineNo">471</span> validateFamiliesInHFiles(table, queue, silence);<a name="line.471"></a>
+<span class="sourceLineNo">472</span> }<a name="line.472"></a>
+<span class="sourceLineNo">473</span><a name="line.473"></a>
+<span class="sourceLineNo">474</span> // Initialize a thread pool<a name="line.474"></a>
+<span class="sourceLineNo">475</span> private ExecutorService createExecutorService() {<a name="line.475"></a>
+<span class="sourceLineNo">476</span> ThreadFactoryBuilder builder = new ThreadFactoryBuilder();<a name="line.476"></a>
+<span class="sourceLineNo">477</span> builder.setNameFormat("LoadIncrementalHFiles-%1$d");<a name="line.477"></a>
+<span class="sourceLineNo">478</span> ExecutorService pool = new ThreadPoolExecutor(nrThreads, nrThreads, 60, TimeUnit.SECONDS,<a name="line.478"></a>
+<span class="sourceLineNo">479</span> new LinkedBlockingQueue<Runnable>(), builder.build());<a name="line.479"></a>
+<span class="sourceLineNo">480</span> ((ThreadPoolExecutor) pool).allowCoreThreadTimeOut(true);<a name="line.480"></a>
+<span class="sourceLineNo">481</span> return pool;<a name="line.481"></a>
+<span class="sourceLineNo">482</span> }<a name="line.482"></a>
+<span class="sourceLineNo">483</span><a name="line.483"></a>
+<span class="sourceLineNo">484</span> /**<a name="line.484"></a>
+<span class="sourceLineNo">485</span> * Checks whether there is any invalid family name in HFiles to be bulk loaded.<a name="line.485"></a>
<span class="sourceLineNo">486</span> */<a name="line.486"></a>
-<span class="sourceLineNo">487</span> public void loadHFileQueue(final Table table, final Connection conn, Deque<LoadQueueItem> queue,<a name="line.487"></a>
-<span class="sourceLineNo">488</span> Pair<byte[][], byte[][]> startEndKeys) throws IOException {<a name="line.488"></a>
-<span class="sourceLineNo">489</span> ExecutorService pool = null;<a name="line.489"></a>
-<span class="sourceLineNo">490</span> try {<a name="line.490"></a>
-<span class="sourceLineNo">491</span> pool = createExecutorService();<a name="line.491"></a>
-<span class="sourceLineNo">492</span> Multimap<ByteBuffer, LoadQueueItem> regionGroups =<a name="line.492"></a>
-<span class="sourceLineNo">493</span> groupOrSplitPhase(table, pool, queue, startEndKeys);<a name="line.493"></a>
-<span class="sourceLineNo">494</span> bulkLoadPhase(table, conn, pool, queue, regionGroups);<a name="line.494"></a>
-<span class="sourceLineNo">495</span> } finally {<a name="line.495"></a>
-<span class="sourceLineNo">496</span> if (pool != null) {<a name="line.496"></a>
-<span class="sourceLineNo">497</span> pool.shutdown();<a name="line.497"></a>
-<span class="sourceLineNo">498</span> }<a name="line.498"></a>
-<span class="sourceLineNo">499</span> }<a name="line.499"></a>
-<span class="sourceLineNo">500</span> }<a name="line.500"></a>
-<span class="sourceLineNo">501</span><a name="line.501"></a>
-<span class="sourceLineNo">502</span> /**<a name="line.502"></a>
-<span class="sourceLineNo">503</span> * This takes the LQI's grouped by likely regions and attempts to bulk load<a name="line.503"></a>
-<span class="sourceLineNo">504</span> * them. Any failures are re-queued for another pass with the<a name="line.504"></a>
-<span class="sourceLineNo">505</span> * groupOrSplitPhase.<a name="line.505"></a>
-<span class="sourceLineNo">506</span> */<a name="line.506"></a>
-<span class="sourceLineNo">507</span> protected void bulkLoadPhase(final Table table, final Connection conn,<a name="line.507"></a>
-<span class="sourceLineNo">508</span> ExecutorService pool, Deque<LoadQueueItem> queue,<a name="line.508"></a>
-<span class="sourceLineNo">509</span> final Multimap<ByteBuffer, LoadQueueItem> regionGroups) throws IOException {<a name="line.509"></a>
-<span class="sourceLineNo">510</span> // atomically bulk load the groups.<a name="line.510"></a>
-<span class="sourceLineNo">511</span> Set<Future<List<LoadQueueItem>>> loadingFutures = new HashSet<>();<a name="line.511"></a>
-<span class="sourceLineNo">512</span> for (Entry<ByteBuffer, ? extends Collection<LoadQueueItem>> e: regionGroups.asMap().entrySet()){<a name="line.512"></a>
-<span class="sourceLineNo">513</span> final byte[] first = e.getKey().array();<a name="line.513"></a>
-<span class="sourceLineNo">514</span> final Collection<LoadQueueItem> lqis = e.getValue();<a name="line.514"></a>
-<span class="sourceLineNo">515</span><a name="line.515"></a>
-<span class="sourceLineNo">516</span> final Callable<List<LoadQueueItem>> call = new Callable<List<LoadQueueItem>>() {<a name="line.516"></a>
-<span class="sourceLineNo">517</span> @Override<a name="line.517"></a>
-<span class="sourceLineNo">518</span> public List<LoadQueueItem> call() throws Exception {<a name="line.518"></a>
-<span class="sourceLineNo">519</span> List<LoadQueueItem> toRetry =<a name="line.519"></a>
-<span class="sourceLineNo">520</span> tryAtomicRegionLoad(conn, table.getName(), first, lqis);<a name="line.520"></a>
-<span class="sourceLineNo">521</span> return toRetry;<a name="line.521"></a>
-<span class="sourceLineNo">522</span> }<a name="line.522"></a>
-<span class="sourceLineNo">523</span> };<a name="line.523"></a>
-<span class="sourceLineNo">524</span> loadingFutures.add(pool.submit(call));<a name="line.524"></a>
-<span class="sourceLineNo">525</span> }<a name="line.525"></a>
-<span class="sourceLineNo">526</span><a name="line.526"></a>
-<span class="sourceLineNo">527</span> // get all the results.<a name="line.527"></a>
-<span class="sourceLineNo">528</span> for (Future<List<LoadQueueItem>> future : loadingFutures) {<a name="line.528"></a>
-<span class="sourceLineNo">529</span> try {<a name="line.529"></a>
-<span class="sourceLineNo">530</span> List<LoadQueueItem> toRetry = future.get();<a name="line.530"></a>
-<span class="sourceLineNo">531</span><a name="line.531"></a>
-<span class="sourceLineNo">532</span> // LQIs that are requeued to be regrouped.<a name="line.532"></a>
-<span class="sourceLineNo">533</span> queue.addAll(toRetry);<a name="line.533"></a>
-<span class="sourceLineNo">534</span><a name="line.534"></a>
-<span class="sourceLineNo">535</span> } catch (ExecutionException e1) {<a name="line.535"></a>
-<span class="sourceLineNo">536</span> Throwable t = e1.getCause();<a name="line.536"></a>
-<span class="sourceLineNo">537</span> if (t instanceof IOException) {<a name="line.537"></a>
-<span class="sourceLineNo">538</span> // At this point something unrecoverable has happened.<a name="line.538"></a>
-<span class="sourceLineNo">539</span> // TODO Implement bulk load recovery<a name="line.539"></a>
-<span class="sourceLineNo">540</span> throw new IOException("BulkLoad encountered an unrecoverable problem", t);<a name="line.540"></a>
-<span class="sourceLineNo">541</span> }<a name="line.541"></a>
-<span class="sourceLineNo">542</span> LOG.error("Unexpected execution exception during bulk load", e1);<a name="line.542"></a>
-<span class="sourceLineNo">543</span> throw new IllegalStateException(t);<a name="line.543"></a>
-<span class="sourceLineNo">544</span> } catch (InterruptedException e1) {<a name="line.544"></a>
-<span class="sourceLineNo">545</span> LOG.error("Unexpected interrupted exception during bulk load", e1);<a name="line.545"></a>
-<span class="sourceLineNo">546</span> throw (InterruptedIOException)new InterruptedIOException().initCause(e1);<a name="line.546"></a>
-<span class="sourceLineNo">547</span> }<a name="line.547"></a>
-<span class="sourceLineNo">548</span> }<a name="line.548"></a>
-<span class="sourceLineNo">549</span> }<a name="line.549"></a>
-<span class="sourceLineNo">550</span><a name="line.550"></a>
-<span class="sourceLineNo">551</span> private boolean checkHFilesCountPerRegionPerFamily(<a name="line.551"></a>
-<span class="sourceLineNo">552</span> final Multimap<ByteBuffer, LoadQueueItem> regionGroups) {<a name="line.552"></a>
-<span class="sourceLineNo">553</span> for (Entry<ByteBuffer,<a name="line.553"></a>
-<span class="sourceLineNo">554</span> ? extends Collection<LoadQueueItem>> e: regionGroups.asMap().entrySet()) {<a name="line.554"></a>
-<span class="sourceLineNo">555</span> final Collection<LoadQueueItem> lqis = e.getValue();<a name="line.555"></a>
-<span class="sourceLineNo">556</span> HashMap<byte[], MutableInt> filesMap = new HashMap<>();<a name="line.556"></a>
-<span class="sourceLineNo">557</span> for (LoadQueueItem lqi: lqis) {<a name="line.557"></a>
-<span class="sourceLineNo">558</span> MutableInt count = filesMap.get(lqi.family);<a name="line.558"></a>
-<span class="sourceLineNo">559</span> if (count == null) {<a name="line.559"></a>
-<span class="sourceLineNo">560</span> count = new MutableInt();<a name="line.560"></a>
-<span class="sourceLineNo">561</span> filesMap.put(lqi.family, count);<a name="line.561"></a>
-<span class="sourceLineNo">562</span> }<a name="line.562"></a>
-<span class="sourceLineNo">563</span> count.increment();<a name="line.563"></a>
-<span class="sourceLineNo">564</span> if (count.intValue() > maxFilesPerRegionPerFamily) {<a name="line.564"></a>
-<span class="sourceLineNo">565</span> LOG.error("Trying to load more than " + maxFilesPerRegionPerFamily<a name="line.565"></a>
-<span class="sourceLineNo">566</span> + " hfiles to family " + Bytes.toStringBinary(lqi.family)<a name="line.566"></a>
-<span class="sourceLineNo">567</span> + " of region with start key "<a name="line.567"></a>
-<span class="sourceLineNo">568</span> + Bytes.toStringBinary(e.getKey()));<a name="line.568"></a>
-<span class="sourceLineNo">569</span> return false;<a name="line.569"></a>
-<span class="sourceLineNo">570</span> }<a name="line.570"></a>
-<span class="sourceLineNo">571</span> }<a name="line.571"></a>
-<span class="sourceLineNo">572</span> }<a name="line.572"></a>
-<span class="sourceLineNo">573</span> return true;<a name="line.573"></a>
-<span class="sourceLineNo">574</span> }<a name="line.574"></a>
-<span class="sourceLineNo">575</span><a name="line.575"></a>
-<span class="sourceLineNo">576</span> /**<a name="line.576"></a>
-<span class="sourceLineNo">577</span> * @return A map that groups LQI by likely bulk load region targets.<a name="line.577"></a>
-<span class="sourceLineNo">578</span> */<a name="line.578"></a>
-<span class="sourceLineNo">579</span> private Multimap<ByteBuffer, LoadQueueItem> groupOrSplitPhase(final Table table,<a name="line.579"></a>
-<span class="sourceLineNo">580</span> ExecutorService pool, Deque<LoadQueueItem> queue,<a name="line.580"></a>
-<span class="sourceLineNo">581</span> final Pair<byte[][], byte[][]> startEndKeys) throws IOException {<a name="line.581"></a>
-<span class="sourceLineNo">582</span> // <region start key, LQI> need synchronized only within this scope of this<a name="line.582"></a>
-<span class="sourceLineNo">583</span> // phase because of the puts that happen in futures.<a name="line.583"></a>
-<span class="sourceLineNo">584</span> Multimap<ByteBuffer, LoadQueueItem> rgs = HashMultimap.create();<a name="line.584"></a>
-<span class="sourceLineNo">585</span> final Multimap<ByteBuffer, LoadQueueItem> regionGroups = Multimaps.synchronizedMultimap(rgs);<a name="line.585"></a>
-<span class="sourceLineNo">586</span><a name="line.586"></a>
-<span class="sourceLineNo">587</span> // drain LQIs and figure out bulk load groups<a name="line.587"></a>
-<span class="sourceLineNo">588</span> Set<Future<List<LoadQueueItem>>> splittingFutures = new HashSet<>();<a name="line.588"></a>
-<span class="sourceLineNo">589</span> while (!queue.isEmpty()) {<a name="line.589"></a>
-<span class="sourceLineNo">590</span> final LoadQueueItem item = queue.remove();<a name="line.590"></a>
-<span class="sourceLineNo">591</span><a name="line.591"></a>
-<span class="sourceLineNo">592</span> final Callable<List<LoadQueueItem>> call = new Callable<List<LoadQueueItem>>() {<a name="line.592"></a>
-<span class="sourceLineNo">593</span> @Override<a name="line.593"></a>
-<span class="sourceLineNo">594</span> public List<LoadQueueItem> call() throws Exception {<a name="line.594"></a>
-<span class="sourceLineNo">595</span> List<LoadQueueItem> splits = groupOrSplit(regionGroups, item, table, startEndKeys);<a name="line.595"></a>
-<span class="sourceLineNo">596</span> return splits;<a name="line.596"></a>
-<span class="sourceLineNo">597</span> }<a name="line.597"></a>
-<span class="sourceLineNo">598</span> };<a name="line.598"></a>
-<span class="sourceLineNo">599</span> splittingFutures.add(pool.submit(call));<a name="line.599"></a>
-<span class="sourceLineNo">600</span> }<a name="line.600"></a>
-<span class="sourceLineNo">601</span> // get all the results. All grouping and splitting must finish before<a name="line.601"></a>
-<span class="sourceLineNo">602</span> // we can attempt the atomic loads.<a name="line.602"></a>
-<span class="sourceLineNo">603</span> for (Future<List<LoadQueueItem>> lqis : splittingFutures) {<a name="line.603"></a>
-<span class="sourceLineNo">604</span> try {<a name="line.604"></a>
-<span class="sourceLineNo">605</span> List<LoadQueueItem> splits = lqis.get();<a name="line.605"></a>
-<span class="sourceLineNo">606</span> if (splits != null) {<a name="line.606"></a>
-<span class="sourceLineNo">607</span> queue.addAll(splits);<a name="line.607"></a>
-<span class="sourceLineNo">608</span> }<a name="line.608"></a>
-<span class="sourceLineNo">609</span> } catch (ExecutionException e1) {<a name="line.609"></a>
-<span class="sourceLineNo">610</span> Throwable t = e1.getCause();<a name="line.610"></a>
-<span class="sourceLineNo">611</span> if (t instanceof IOException) {<a name="line.611"></a>
-<span class="sourceLineNo">612</span> LOG.error("IOException during splitting", e1);<a name="line.612"></a>
-<span class="sourceLineNo">613</span> throw (IOException)t; // would have been thrown if not parallelized,<a name="line.613"></a>
-<span class="sourceLineNo">614</span> }<a name="line.614"></a>
-<span class="sourceLineNo">615</span> LOG.error("Unexpected execution exception during splitting", e1);<a name="line.615"></a>
-<span class="sourceLineNo">616</span> throw new IllegalStateException(t);<a name="line.616"></a>
-<span class="sourceLineNo">617</span> } catch (InterruptedException e1) {<a name="line.617"></a>
-<span class="sourceLineNo">618</span> LOG.error("Unexpected interrupted exception during splitting", e1);<a name="line.618"></a>
-<span class="sourceLineNo">619</span> throw (InterruptedI
<TRUNCATED>