You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by gi...@apache.org on 2017/05/26 15:00:10 UTC
[20/32] hbase-site git commit: Published site at
82d554e3783372cc6b05489452c815b57c06f6cd.
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/be6f38d2/devapidocs/src-html/org/apache/hadoop/hbase/client/TableSnapshotScanner.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/client/TableSnapshotScanner.html b/devapidocs/src-html/org/apache/hadoop/hbase/client/TableSnapshotScanner.html
index 5566e01..9615ff3 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/client/TableSnapshotScanner.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/client/TableSnapshotScanner.html
@@ -134,68 +134,71 @@
<span class="sourceLineNo">126</span><a name="line.126"></a>
<span class="sourceLineNo">127</span> htd = meta.getTableDescriptor();<a name="line.127"></a>
<span class="sourceLineNo">128</span> regions = new ArrayList<>(restoredRegions.size());<a name="line.128"></a>
-<span class="sourceLineNo">129</span> for (HRegionInfo hri: restoredRegions) {<a name="line.129"></a>
-<span class="sourceLineNo">130</span> if (CellUtil.overlappingKeys(scan.getStartRow(), scan.getStopRow(),<a name="line.130"></a>
-<span class="sourceLineNo">131</span> hri.getStartKey(), hri.getEndKey())) {<a name="line.131"></a>
-<span class="sourceLineNo">132</span> regions.add(hri);<a name="line.132"></a>
-<span class="sourceLineNo">133</span> }<a name="line.133"></a>
-<span class="sourceLineNo">134</span> }<a name="line.134"></a>
-<span class="sourceLineNo">135</span><a name="line.135"></a>
-<span class="sourceLineNo">136</span> // sort for regions according to startKey.<a name="line.136"></a>
-<span class="sourceLineNo">137</span> Collections.sort(regions);<a name="line.137"></a>
-<span class="sourceLineNo">138</span> initScanMetrics(scan);<a name="line.138"></a>
-<span class="sourceLineNo">139</span> }<a name="line.139"></a>
-<span class="sourceLineNo">140</span><a name="line.140"></a>
-<span class="sourceLineNo">141</span> @Override<a name="line.141"></a>
-<span class="sourceLineNo">142</span> public Result next() throws IOException {<a name="line.142"></a>
-<span class="sourceLineNo">143</span> Result result = null;<a name="line.143"></a>
-<span class="sourceLineNo">144</span> while (true) {<a name="line.144"></a>
-<span class="sourceLineNo">145</span> if (currentRegionScanner == null) {<a name="line.145"></a>
-<span class="sourceLineNo">146</span> currentRegion++;<a name="line.146"></a>
-<span class="sourceLineNo">147</span> if (currentRegion >= regions.size()) {<a name="line.147"></a>
-<span class="sourceLineNo">148</span> return null;<a name="line.148"></a>
-<span class="sourceLineNo">149</span> }<a name="line.149"></a>
-<span class="sourceLineNo">150</span><a name="line.150"></a>
-<span class="sourceLineNo">151</span> HRegionInfo hri = regions.get(currentRegion);<a name="line.151"></a>
-<span class="sourceLineNo">152</span> currentRegionScanner = new ClientSideRegionScanner(conf, fs,<a name="line.152"></a>
-<span class="sourceLineNo">153</span> restoreDir, htd, hri, scan, scanMetrics);<a name="line.153"></a>
-<span class="sourceLineNo">154</span> if (this.scanMetrics != null) {<a name="line.154"></a>
-<span class="sourceLineNo">155</span> this.scanMetrics.countOfRegions.incrementAndGet();<a name="line.155"></a>
-<span class="sourceLineNo">156</span> }<a name="line.156"></a>
-<span class="sourceLineNo">157</span> }<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span> try {<a name="line.159"></a>
-<span class="sourceLineNo">160</span> result = currentRegionScanner.next();<a name="line.160"></a>
-<span class="sourceLineNo">161</span> if (result != null) {<a name="line.161"></a>
-<span class="sourceLineNo">162</span> return result;<a name="line.162"></a>
-<span class="sourceLineNo">163</span> }<a name="line.163"></a>
-<span class="sourceLineNo">164</span> } finally {<a name="line.164"></a>
-<span class="sourceLineNo">165</span> if (result == null) {<a name="line.165"></a>
-<span class="sourceLineNo">166</span> currentRegionScanner.close();<a name="line.166"></a>
-<span class="sourceLineNo">167</span> currentRegionScanner = null;<a name="line.167"></a>
-<span class="sourceLineNo">168</span> }<a name="line.168"></a>
-<span class="sourceLineNo">169</span> }<a name="line.169"></a>
-<span class="sourceLineNo">170</span> }<a name="line.170"></a>
-<span class="sourceLineNo">171</span> }<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> @Override<a name="line.173"></a>
-<span class="sourceLineNo">174</span> public void close() {<a name="line.174"></a>
-<span class="sourceLineNo">175</span> if (currentRegionScanner != null) {<a name="line.175"></a>
-<span class="sourceLineNo">176</span> currentRegionScanner.close();<a name="line.176"></a>
-<span class="sourceLineNo">177</span> }<a name="line.177"></a>
-<span class="sourceLineNo">178</span> try {<a name="line.178"></a>
-<span class="sourceLineNo">179</span> fs.delete(this.restoreDir, true);<a name="line.179"></a>
-<span class="sourceLineNo">180</span> } catch (IOException ex) {<a name="line.180"></a>
-<span class="sourceLineNo">181</span> LOG.warn("Could not delete restore directory for the snapshot:" + ex);<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><a name="line.184"></a>
-<span class="sourceLineNo">185</span> @Override<a name="line.185"></a>
-<span class="sourceLineNo">186</span> public boolean renewLease() {<a name="line.186"></a>
-<span class="sourceLineNo">187</span> throw new UnsupportedOperationException();<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>}<a name="line.190"></a>
+<span class="sourceLineNo">129</span> for (HRegionInfo hri : restoredRegions) {<a name="line.129"></a>
+<span class="sourceLineNo">130</span> if (hri.isOffline() && (hri.isSplit() || hri.isSplitParent())) {<a name="line.130"></a>
+<span class="sourceLineNo">131</span> continue;<a name="line.131"></a>
+<span class="sourceLineNo">132</span> }<a name="line.132"></a>
+<span class="sourceLineNo">133</span> if (CellUtil.overlappingKeys(scan.getStartRow(), scan.getStopRow(), hri.getStartKey(),<a name="line.133"></a>
+<span class="sourceLineNo">134</span> hri.getEndKey())) {<a name="line.134"></a>
+<span class="sourceLineNo">135</span> regions.add(hri);<a name="line.135"></a>
+<span class="sourceLineNo">136</span> }<a name="line.136"></a>
+<span class="sourceLineNo">137</span> }<a name="line.137"></a>
+<span class="sourceLineNo">138</span><a name="line.138"></a>
+<span class="sourceLineNo">139</span> // sort for regions according to startKey.<a name="line.139"></a>
+<span class="sourceLineNo">140</span> Collections.sort(regions);<a name="line.140"></a>
+<span class="sourceLineNo">141</span> initScanMetrics(scan);<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> @Override<a name="line.144"></a>
+<span class="sourceLineNo">145</span> public Result next() throws IOException {<a name="line.145"></a>
+<span class="sourceLineNo">146</span> Result result = null;<a name="line.146"></a>
+<span class="sourceLineNo">147</span> while (true) {<a name="line.147"></a>
+<span class="sourceLineNo">148</span> if (currentRegionScanner == null) {<a name="line.148"></a>
+<span class="sourceLineNo">149</span> currentRegion++;<a name="line.149"></a>
+<span class="sourceLineNo">150</span> if (currentRegion >= regions.size()) {<a name="line.150"></a>
+<span class="sourceLineNo">151</span> return null;<a name="line.151"></a>
+<span class="sourceLineNo">152</span> }<a name="line.152"></a>
+<span class="sourceLineNo">153</span><a name="line.153"></a>
+<span class="sourceLineNo">154</span> HRegionInfo hri = regions.get(currentRegion);<a name="line.154"></a>
+<span class="sourceLineNo">155</span> currentRegionScanner = new ClientSideRegionScanner(conf, fs,<a name="line.155"></a>
+<span class="sourceLineNo">156</span> restoreDir, htd, hri, scan, scanMetrics);<a name="line.156"></a>
+<span class="sourceLineNo">157</span> if (this.scanMetrics != null) {<a name="line.157"></a>
+<span class="sourceLineNo">158</span> this.scanMetrics.countOfRegions.incrementAndGet();<a name="line.158"></a>
+<span class="sourceLineNo">159</span> }<a name="line.159"></a>
+<span class="sourceLineNo">160</span> }<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> try {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> result = currentRegionScanner.next();<a name="line.163"></a>
+<span class="sourceLineNo">164</span> if (result != null) {<a name="line.164"></a>
+<span class="sourceLineNo">165</span> return result;<a name="line.165"></a>
+<span class="sourceLineNo">166</span> }<a name="line.166"></a>
+<span class="sourceLineNo">167</span> } finally {<a name="line.167"></a>
+<span class="sourceLineNo">168</span> if (result == null) {<a name="line.168"></a>
+<span class="sourceLineNo">169</span> currentRegionScanner.close();<a name="line.169"></a>
+<span class="sourceLineNo">170</span> currentRegionScanner = null;<a name="line.170"></a>
+<span class="sourceLineNo">171</span> }<a name="line.171"></a>
+<span class="sourceLineNo">172</span> }<a name="line.172"></a>
+<span class="sourceLineNo">173</span> }<a name="line.173"></a>
+<span class="sourceLineNo">174</span> }<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span> @Override<a name="line.176"></a>
+<span class="sourceLineNo">177</span> public void close() {<a name="line.177"></a>
+<span class="sourceLineNo">178</span> if (currentRegionScanner != null) {<a name="line.178"></a>
+<span class="sourceLineNo">179</span> currentRegionScanner.close();<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span> try {<a name="line.181"></a>
+<span class="sourceLineNo">182</span> fs.delete(this.restoreDir, true);<a name="line.182"></a>
+<span class="sourceLineNo">183</span> } catch (IOException ex) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span> LOG.warn("Could not delete restore directory for the snapshot:" + ex);<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><a name="line.187"></a>
+<span class="sourceLineNo">188</span> @Override<a name="line.188"></a>
+<span class="sourceLineNo">189</span> public boolean renewLease() {<a name="line.189"></a>
+<span class="sourceLineNo">190</span> throw new UnsupportedOperationException();<a name="line.190"></a>
+<span class="sourceLineNo">191</span> }<a name="line.191"></a>
+<span class="sourceLineNo">192</span><a name="line.192"></a>
+<span class="sourceLineNo">193</span>}<a name="line.193"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/be6f38d2/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.InputSplit.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.InputSplit.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.InputSplit.html
index 837c395..2578d72 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.InputSplit.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.InputSplit.html
@@ -283,137 +283,141 @@
<span class="sourceLineNo">275</span> List<HRegionInfo> regionInfos = Lists.newArrayListWithCapacity(regionManifests.size());<a name="line.275"></a>
<span class="sourceLineNo">276</span><a name="line.276"></a>
<span class="sourceLineNo">277</span> for (SnapshotRegionManifest regionManifest : regionManifests) {<a name="line.277"></a>
-<span class="sourceLineNo">278</span> regionInfos.add(HRegionInfo.convert(regionManifest.getRegionInfo()));<a name="line.278"></a>
-<span class="sourceLineNo">279</span> }<a name="line.279"></a>
-<span class="sourceLineNo">280</span> return regionInfos;<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> public static SnapshotManifest getSnapshotManifest(Configuration conf, String snapshotName,<a name="line.283"></a>
-<span class="sourceLineNo">284</span> Path rootDir, FileSystem fs) throws IOException {<a name="line.284"></a>
-<span class="sourceLineNo">285</span> Path snapshotDir = SnapshotDescriptionUtils.getCompletedSnapshotDir(snapshotName, rootDir);<a name="line.285"></a>
-<span class="sourceLineNo">286</span> SnapshotDescription snapshotDesc = SnapshotDescriptionUtils.readSnapshotInfo(fs, snapshotDir);<a name="line.286"></a>
-<span class="sourceLineNo">287</span> return SnapshotManifest.open(conf, fs, snapshotDir, snapshotDesc);<a name="line.287"></a>
-<span class="sourceLineNo">288</span> }<a name="line.288"></a>
-<span class="sourceLineNo">289</span><a name="line.289"></a>
-<span class="sourceLineNo">290</span> public static Scan extractScanFromConf(Configuration conf) throws IOException {<a name="line.290"></a>
-<span class="sourceLineNo">291</span> Scan scan = null;<a name="line.291"></a>
-<span class="sourceLineNo">292</span> if (conf.get(TableInputFormat.SCAN) != null) {<a name="line.292"></a>
-<span class="sourceLineNo">293</span> scan = TableMapReduceUtil.convertStringToScan(conf.get(TableInputFormat.SCAN));<a name="line.293"></a>
-<span class="sourceLineNo">294</span> } else if (conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST) != null) {<a name="line.294"></a>
-<span class="sourceLineNo">295</span> String[] columns =<a name="line.295"></a>
-<span class="sourceLineNo">296</span> conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST).split(" ");<a name="line.296"></a>
-<span class="sourceLineNo">297</span> scan = new Scan();<a name="line.297"></a>
-<span class="sourceLineNo">298</span> for (String col : columns) {<a name="line.298"></a>
-<span class="sourceLineNo">299</span> scan.addFamily(Bytes.toBytes(col));<a name="line.299"></a>
-<span class="sourceLineNo">300</span> }<a name="line.300"></a>
-<span class="sourceLineNo">301</span> } else {<a name="line.301"></a>
-<span class="sourceLineNo">302</span> throw new IllegalArgumentException("Unable to create scan");<a name="line.302"></a>
-<span class="sourceLineNo">303</span> }<a name="line.303"></a>
-<span class="sourceLineNo">304</span> return scan;<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> public static List<InputSplit> getSplits(Scan scan, SnapshotManifest manifest,<a name="line.307"></a>
-<span class="sourceLineNo">308</span> List<HRegionInfo> regionManifests, Path restoreDir, Configuration conf) throws IOException {<a name="line.308"></a>
-<span class="sourceLineNo">309</span> // load table descriptor<a name="line.309"></a>
-<span class="sourceLineNo">310</span> HTableDescriptor htd = manifest.getTableDescriptor();<a name="line.310"></a>
-<span class="sourceLineNo">311</span><a name="line.311"></a>
-<span class="sourceLineNo">312</span> Path tableDir = FSUtils.getTableDir(restoreDir, htd.getTableName());<a name="line.312"></a>
-<span class="sourceLineNo">313</span><a name="line.313"></a>
-<span class="sourceLineNo">314</span> List<InputSplit> splits = new ArrayList<>();<a name="line.314"></a>
-<span class="sourceLineNo">315</span> for (HRegionInfo hri : regionManifests) {<a name="line.315"></a>
-<span class="sourceLineNo">316</span> // load region descriptor<a name="line.316"></a>
+<span class="sourceLineNo">278</span> HRegionInfo hri = HRegionInfo.convert(regionManifest.getRegionInfo());<a name="line.278"></a>
+<span class="sourceLineNo">279</span> if (hri.isOffline() && (hri.isSplit() || hri.isSplitParent())) {<a name="line.279"></a>
+<span class="sourceLineNo">280</span> continue;<a name="line.280"></a>
+<span class="sourceLineNo">281</span> }<a name="line.281"></a>
+<span class="sourceLineNo">282</span> regionInfos.add(hri);<a name="line.282"></a>
+<span class="sourceLineNo">283</span> }<a name="line.283"></a>
+<span class="sourceLineNo">284</span> return regionInfos;<a name="line.284"></a>
+<span class="sourceLineNo">285</span> }<a name="line.285"></a>
+<span class="sourceLineNo">286</span><a name="line.286"></a>
+<span class="sourceLineNo">287</span> public static SnapshotManifest getSnapshotManifest(Configuration conf, String snapshotName,<a name="line.287"></a>
+<span class="sourceLineNo">288</span> Path rootDir, FileSystem fs) throws IOException {<a name="line.288"></a>
+<span class="sourceLineNo">289</span> Path snapshotDir = SnapshotDescriptionUtils.getCompletedSnapshotDir(snapshotName, rootDir);<a name="line.289"></a>
+<span class="sourceLineNo">290</span> SnapshotDescription snapshotDesc = SnapshotDescriptionUtils.readSnapshotInfo(fs, snapshotDir);<a name="line.290"></a>
+<span class="sourceLineNo">291</span> return SnapshotManifest.open(conf, fs, snapshotDir, snapshotDesc);<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> public static Scan extractScanFromConf(Configuration conf) throws IOException {<a name="line.294"></a>
+<span class="sourceLineNo">295</span> Scan scan = null;<a name="line.295"></a>
+<span class="sourceLineNo">296</span> if (conf.get(TableInputFormat.SCAN) != null) {<a name="line.296"></a>
+<span class="sourceLineNo">297</span> scan = TableMapReduceUtil.convertStringToScan(conf.get(TableInputFormat.SCAN));<a name="line.297"></a>
+<span class="sourceLineNo">298</span> } else if (conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST) != null) {<a name="line.298"></a>
+<span class="sourceLineNo">299</span> String[] columns =<a name="line.299"></a>
+<span class="sourceLineNo">300</span> conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST).split(" ");<a name="line.300"></a>
+<span class="sourceLineNo">301</span> scan = new Scan();<a name="line.301"></a>
+<span class="sourceLineNo">302</span> for (String col : columns) {<a name="line.302"></a>
+<span class="sourceLineNo">303</span> scan.addFamily(Bytes.toBytes(col));<a name="line.303"></a>
+<span class="sourceLineNo">304</span> }<a name="line.304"></a>
+<span class="sourceLineNo">305</span> } else {<a name="line.305"></a>
+<span class="sourceLineNo">306</span> throw new IllegalArgumentException("Unable to create scan");<a name="line.306"></a>
+<span class="sourceLineNo">307</span> }<a name="line.307"></a>
+<span class="sourceLineNo">308</span> return scan;<a name="line.308"></a>
+<span class="sourceLineNo">309</span> }<a name="line.309"></a>
+<span class="sourceLineNo">310</span><a name="line.310"></a>
+<span class="sourceLineNo">311</span> public static List<InputSplit> getSplits(Scan scan, SnapshotManifest manifest,<a name="line.311"></a>
+<span class="sourceLineNo">312</span> List<HRegionInfo> regionManifests, Path restoreDir, Configuration conf) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> // load table descriptor<a name="line.313"></a>
+<span class="sourceLineNo">314</span> HTableDescriptor htd = manifest.getTableDescriptor();<a name="line.314"></a>
+<span class="sourceLineNo">315</span><a name="line.315"></a>
+<span class="sourceLineNo">316</span> Path tableDir = FSUtils.getTableDir(restoreDir, htd.getTableName());<a name="line.316"></a>
<span class="sourceLineNo">317</span><a name="line.317"></a>
-<span class="sourceLineNo">318</span> if (CellUtil.overlappingKeys(scan.getStartRow(), scan.getStopRow(), hri.getStartKey(),<a name="line.318"></a>
-<span class="sourceLineNo">319</span> hri.getEndKey())) {<a name="line.319"></a>
-<span class="sourceLineNo">320</span> // compute HDFS locations from snapshot files (which will get the locations for<a name="line.320"></a>
-<span class="sourceLineNo">321</span> // referred hfiles)<a name="line.321"></a>
-<span class="sourceLineNo">322</span> List<String> hosts = getBestLocations(conf,<a name="line.322"></a>
-<span class="sourceLineNo">323</span> HRegion.computeHDFSBlocksDistribution(conf, htd, hri, tableDir));<a name="line.323"></a>
-<span class="sourceLineNo">324</span><a name="line.324"></a>
-<span class="sourceLineNo">325</span> int len = Math.min(3, hosts.size());<a name="line.325"></a>
-<span class="sourceLineNo">326</span> hosts = hosts.subList(0, len);<a name="line.326"></a>
-<span class="sourceLineNo">327</span> splits.add(new InputSplit(htd, hri, hosts, scan, restoreDir));<a name="line.327"></a>
-<span class="sourceLineNo">328</span> }<a name="line.328"></a>
-<span class="sourceLineNo">329</span> }<a name="line.329"></a>
-<span class="sourceLineNo">330</span><a name="line.330"></a>
-<span class="sourceLineNo">331</span> return splits;<a name="line.331"></a>
-<span class="sourceLineNo">332</span><a name="line.332"></a>
-<span class="sourceLineNo">333</span> }<a name="line.333"></a>
+<span class="sourceLineNo">318</span> List<InputSplit> splits = new ArrayList<>();<a name="line.318"></a>
+<span class="sourceLineNo">319</span> for (HRegionInfo hri : regionManifests) {<a name="line.319"></a>
+<span class="sourceLineNo">320</span> // load region descriptor<a name="line.320"></a>
+<span class="sourceLineNo">321</span><a name="line.321"></a>
+<span class="sourceLineNo">322</span> if (CellUtil.overlappingKeys(scan.getStartRow(), scan.getStopRow(), hri.getStartKey(),<a name="line.322"></a>
+<span class="sourceLineNo">323</span> hri.getEndKey())) {<a name="line.323"></a>
+<span class="sourceLineNo">324</span> // compute HDFS locations from snapshot files (which will get the locations for<a name="line.324"></a>
+<span class="sourceLineNo">325</span> // referred hfiles)<a name="line.325"></a>
+<span class="sourceLineNo">326</span> List<String> hosts = getBestLocations(conf,<a name="line.326"></a>
+<span class="sourceLineNo">327</span> HRegion.computeHDFSBlocksDistribution(conf, htd, hri, tableDir));<a name="line.327"></a>
+<span class="sourceLineNo">328</span><a name="line.328"></a>
+<span class="sourceLineNo">329</span> int len = Math.min(3, hosts.size());<a name="line.329"></a>
+<span class="sourceLineNo">330</span> hosts = hosts.subList(0, len);<a name="line.330"></a>
+<span class="sourceLineNo">331</span> splits.add(new InputSplit(htd, hri, hosts, scan, restoreDir));<a name="line.331"></a>
+<span class="sourceLineNo">332</span> }<a name="line.332"></a>
+<span class="sourceLineNo">333</span> }<a name="line.333"></a>
<span class="sourceLineNo">334</span><a name="line.334"></a>
-<span class="sourceLineNo">335</span> /**<a name="line.335"></a>
-<span class="sourceLineNo">336</span> * This computes the locations to be passed from the InputSplit. MR/Yarn schedulers does not take<a name="line.336"></a>
-<span class="sourceLineNo">337</span> * weights into account, thus will treat every location passed from the input split as equal. We<a name="line.337"></a>
-<span class="sourceLineNo">338</span> * do not want to blindly pass all the locations, since we are creating one split per region, and<a name="line.338"></a>
-<span class="sourceLineNo">339</span> * the region's blocks are all distributed throughout the cluster unless favorite node assignment<a name="line.339"></a>
-<span class="sourceLineNo">340</span> * is used. On the expected stable case, only one location will contain most of the blocks as<a name="line.340"></a>
-<span class="sourceLineNo">341</span> * local.<a name="line.341"></a>
-<span class="sourceLineNo">342</span> * On the other hand, in favored node assignment, 3 nodes will contain highly local blocks. Here<a name="line.342"></a>
-<span class="sourceLineNo">343</span> * we are doing a simple heuristic, where we will pass all hosts which have at least 80%<a name="line.343"></a>
-<span class="sourceLineNo">344</span> * (hbase.tablesnapshotinputformat.locality.cutoff.multiplier) as much block locality as the top<a name="line.344"></a>
-<span class="sourceLineNo">345</span> * host with the best locality.<a name="line.345"></a>
-<span class="sourceLineNo">346</span> */<a name="line.346"></a>
-<span class="sourceLineNo">347</span> public static List<String> getBestLocations(<a name="line.347"></a>
-<span class="sourceLineNo">348</span> Configuration conf, HDFSBlocksDistribution blockDistribution) {<a name="line.348"></a>
-<span class="sourceLineNo">349</span> List<String> locations = new ArrayList<>(3);<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> HostAndWeight[] hostAndWeights = blockDistribution.getTopHostsWithWeights();<a name="line.351"></a>
-<span class="sourceLineNo">352</span><a name="line.352"></a>
-<span class="sourceLineNo">353</span> if (hostAndWeights.length == 0) {<a name="line.353"></a>
-<span class="sourceLineNo">354</span> return locations;<a name="line.354"></a>
-<span class="sourceLineNo">355</span> }<a name="line.355"></a>
+<span class="sourceLineNo">335</span> return splits;<a name="line.335"></a>
+<span class="sourceLineNo">336</span><a name="line.336"></a>
+<span class="sourceLineNo">337</span> }<a name="line.337"></a>
+<span class="sourceLineNo">338</span><a name="line.338"></a>
+<span class="sourceLineNo">339</span> /**<a name="line.339"></a>
+<span class="sourceLineNo">340</span> * This computes the locations to be passed from the InputSplit. MR/Yarn schedulers does not take<a name="line.340"></a>
+<span class="sourceLineNo">341</span> * weights into account, thus will treat every location passed from the input split as equal. We<a name="line.341"></a>
+<span class="sourceLineNo">342</span> * do not want to blindly pass all the locations, since we are creating one split per region, and<a name="line.342"></a>
+<span class="sourceLineNo">343</span> * the region's blocks are all distributed throughout the cluster unless favorite node assignment<a name="line.343"></a>
+<span class="sourceLineNo">344</span> * is used. On the expected stable case, only one location will contain most of the blocks as<a name="line.344"></a>
+<span class="sourceLineNo">345</span> * local.<a name="line.345"></a>
+<span class="sourceLineNo">346</span> * On the other hand, in favored node assignment, 3 nodes will contain highly local blocks. Here<a name="line.346"></a>
+<span class="sourceLineNo">347</span> * we are doing a simple heuristic, where we will pass all hosts which have at least 80%<a name="line.347"></a>
+<span class="sourceLineNo">348</span> * (hbase.tablesnapshotinputformat.locality.cutoff.multiplier) as much block locality as the top<a name="line.348"></a>
+<span class="sourceLineNo">349</span> * host with the best locality.<a name="line.349"></a>
+<span class="sourceLineNo">350</span> */<a name="line.350"></a>
+<span class="sourceLineNo">351</span> public static List<String> getBestLocations(<a name="line.351"></a>
+<span class="sourceLineNo">352</span> Configuration conf, HDFSBlocksDistribution blockDistribution) {<a name="line.352"></a>
+<span class="sourceLineNo">353</span> List<String> locations = new ArrayList<>(3);<a name="line.353"></a>
+<span class="sourceLineNo">354</span><a name="line.354"></a>
+<span class="sourceLineNo">355</span> HostAndWeight[] hostAndWeights = blockDistribution.getTopHostsWithWeights();<a name="line.355"></a>
<span class="sourceLineNo">356</span><a name="line.356"></a>
-<span class="sourceLineNo">357</span> HostAndWeight topHost = hostAndWeights[0];<a name="line.357"></a>
-<span class="sourceLineNo">358</span> locations.add(topHost.getHost());<a name="line.358"></a>
-<span class="sourceLineNo">359</span><a name="line.359"></a>
-<span class="sourceLineNo">360</span> // Heuristic: filter all hosts which have at least cutoffMultiplier % of block locality<a name="line.360"></a>
-<span class="sourceLineNo">361</span> double cutoffMultiplier<a name="line.361"></a>
-<span class="sourceLineNo">362</span> = conf.getFloat(LOCALITY_CUTOFF_MULTIPLIER, DEFAULT_LOCALITY_CUTOFF_MULTIPLIER);<a name="line.362"></a>
+<span class="sourceLineNo">357</span> if (hostAndWeights.length == 0) {<a name="line.357"></a>
+<span class="sourceLineNo">358</span> return locations;<a name="line.358"></a>
+<span class="sourceLineNo">359</span> }<a name="line.359"></a>
+<span class="sourceLineNo">360</span><a name="line.360"></a>
+<span class="sourceLineNo">361</span> HostAndWeight topHost = hostAndWeights[0];<a name="line.361"></a>
+<span class="sourceLineNo">362</span> locations.add(topHost.getHost());<a name="line.362"></a>
<span class="sourceLineNo">363</span><a name="line.363"></a>
-<span class="sourceLineNo">364</span> double filterWeight = topHost.getWeight() * cutoffMultiplier;<a name="line.364"></a>
-<span class="sourceLineNo">365</span><a name="line.365"></a>
-<span class="sourceLineNo">366</span> for (int i = 1; i < hostAndWeights.length; i++) {<a name="line.366"></a>
-<span class="sourceLineNo">367</span> if (hostAndWeights[i].getWeight() >= filterWeight) {<a name="line.367"></a>
-<span class="sourceLineNo">368</span> locations.add(hostAndWeights[i].getHost());<a name="line.368"></a>
-<span class="sourceLineNo">369</span> } else {<a name="line.369"></a>
-<span class="sourceLineNo">370</span> break;<a name="line.370"></a>
-<span class="sourceLineNo">371</span> }<a name="line.371"></a>
-<span class="sourceLineNo">372</span> }<a name="line.372"></a>
-<span class="sourceLineNo">373</span><a name="line.373"></a>
-<span class="sourceLineNo">374</span> return locations;<a name="line.374"></a>
-<span class="sourceLineNo">375</span> }<a name="line.375"></a>
-<span class="sourceLineNo">376</span><a name="line.376"></a>
-<span class="sourceLineNo">377</span> private static String getSnapshotName(Configuration conf) {<a name="line.377"></a>
-<span class="sourceLineNo">378</span> String snapshotName = conf.get(SNAPSHOT_NAME_KEY);<a name="line.378"></a>
-<span class="sourceLineNo">379</span> if (snapshotName == null) {<a name="line.379"></a>
-<span class="sourceLineNo">380</span> throw new IllegalArgumentException("Snapshot name must be provided");<a name="line.380"></a>
-<span class="sourceLineNo">381</span> }<a name="line.381"></a>
-<span class="sourceLineNo">382</span> return snapshotName;<a name="line.382"></a>
-<span class="sourceLineNo">383</span> }<a name="line.383"></a>
-<span class="sourceLineNo">384</span><a name="line.384"></a>
-<span class="sourceLineNo">385</span> /**<a name="line.385"></a>
-<span class="sourceLineNo">386</span> * Configures the job to use TableSnapshotInputFormat to read from a snapshot.<a name="line.386"></a>
-<span class="sourceLineNo">387</span> * @param conf the job to configuration<a name="line.387"></a>
-<span class="sourceLineNo">388</span> * @param snapshotName the name of the snapshot to read from<a name="line.388"></a>
-<span class="sourceLineNo">389</span> * @param restoreDir a temporary directory to restore the snapshot into. Current user should<a name="line.389"></a>
-<span class="sourceLineNo">390</span> * have write permissions to this directory, and this should not be a subdirectory of rootdir.<a name="line.390"></a>
-<span class="sourceLineNo">391</span> * After the job is finished, restoreDir can be deleted.<a name="line.391"></a>
-<span class="sourceLineNo">392</span> * @throws IOException if an error occurs<a name="line.392"></a>
-<span class="sourceLineNo">393</span> */<a name="line.393"></a>
-<span class="sourceLineNo">394</span> public static void setInput(Configuration conf, String snapshotName, Path restoreDir)<a name="line.394"></a>
-<span class="sourceLineNo">395</span> throws IOException {<a name="line.395"></a>
-<span class="sourceLineNo">396</span> conf.set(SNAPSHOT_NAME_KEY, snapshotName);<a name="line.396"></a>
-<span class="sourceLineNo">397</span><a name="line.397"></a>
-<span class="sourceLineNo">398</span> Path rootDir = FSUtils.getRootDir(conf);<a name="line.398"></a>
-<span class="sourceLineNo">399</span> FileSystem fs = rootDir.getFileSystem(conf);<a name="line.399"></a>
-<span class="sourceLineNo">400</span><a name="line.400"></a>
-<span class="sourceLineNo">401</span> restoreDir = new Path(restoreDir, UUID.randomUUID().toString());<a name="line.401"></a>
-<span class="sourceLineNo">402</span><a name="line.402"></a>
-<span class="sourceLineNo">403</span> // TODO: restore from record readers to parallelize.<a name="line.403"></a>
-<span class="sourceLineNo">404</span> RestoreSnapshotHelper.copySnapshotForScanner(conf, fs, rootDir, restoreDir, snapshotName);<a name="line.404"></a>
-<span class="sourceLineNo">405</span><a name="line.405"></a>
-<span class="sourceLineNo">406</span> conf.set(RESTORE_DIR_KEY, restoreDir.toString());<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">364</span> // Heuristic: filter all hosts which have at least cutoffMultiplier % of block locality<a name="line.364"></a>
+<span class="sourceLineNo">365</span> double cutoffMultiplier<a name="line.365"></a>
+<span class="sourceLineNo">366</span> = conf.getFloat(LOCALITY_CUTOFF_MULTIPLIER, DEFAULT_LOCALITY_CUTOFF_MULTIPLIER);<a name="line.366"></a>
+<span class="sourceLineNo">367</span><a name="line.367"></a>
+<span class="sourceLineNo">368</span> double filterWeight = topHost.getWeight() * cutoffMultiplier;<a name="line.368"></a>
+<span class="sourceLineNo">369</span><a name="line.369"></a>
+<span class="sourceLineNo">370</span> for (int i = 1; i < hostAndWeights.length; i++) {<a name="line.370"></a>
+<span class="sourceLineNo">371</span> if (hostAndWeights[i].getWeight() >= filterWeight) {<a name="line.371"></a>
+<span class="sourceLineNo">372</span> locations.add(hostAndWeights[i].getHost());<a name="line.372"></a>
+<span class="sourceLineNo">373</span> } else {<a name="line.373"></a>
+<span class="sourceLineNo">374</span> break;<a name="line.374"></a>
+<span class="sourceLineNo">375</span> }<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> return locations;<a name="line.378"></a>
+<span class="sourceLineNo">379</span> }<a name="line.379"></a>
+<span class="sourceLineNo">380</span><a name="line.380"></a>
+<span class="sourceLineNo">381</span> private static String getSnapshotName(Configuration conf) {<a name="line.381"></a>
+<span class="sourceLineNo">382</span> String snapshotName = conf.get(SNAPSHOT_NAME_KEY);<a name="line.382"></a>
+<span class="sourceLineNo">383</span> if (snapshotName == null) {<a name="line.383"></a>
+<span class="sourceLineNo">384</span> throw new IllegalArgumentException("Snapshot name must be provided");<a name="line.384"></a>
+<span class="sourceLineNo">385</span> }<a name="line.385"></a>
+<span class="sourceLineNo">386</span> return snapshotName;<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> /**<a name="line.389"></a>
+<span class="sourceLineNo">390</span> * Configures the job to use TableSnapshotInputFormat to read from a snapshot.<a name="line.390"></a>
+<span class="sourceLineNo">391</span> * @param conf the job to configuration<a name="line.391"></a>
+<span class="sourceLineNo">392</span> * @param snapshotName the name of the snapshot to read from<a name="line.392"></a>
+<span class="sourceLineNo">393</span> * @param restoreDir a temporary directory to restore the snapshot into. Current user should<a name="line.393"></a>
+<span class="sourceLineNo">394</span> * have write permissions to this directory, and this should not be a subdirectory of rootdir.<a name="line.394"></a>
+<span class="sourceLineNo">395</span> * After the job is finished, restoreDir can be deleted.<a name="line.395"></a>
+<span class="sourceLineNo">396</span> * @throws IOException if an error occurs<a name="line.396"></a>
+<span class="sourceLineNo">397</span> */<a name="line.397"></a>
+<span class="sourceLineNo">398</span> public static void setInput(Configuration conf, String snapshotName, Path restoreDir)<a name="line.398"></a>
+<span class="sourceLineNo">399</span> throws IOException {<a name="line.399"></a>
+<span class="sourceLineNo">400</span> conf.set(SNAPSHOT_NAME_KEY, snapshotName);<a name="line.400"></a>
+<span class="sourceLineNo">401</span><a name="line.401"></a>
+<span class="sourceLineNo">402</span> Path rootDir = FSUtils.getRootDir(conf);<a name="line.402"></a>
+<span class="sourceLineNo">403</span> FileSystem fs = rootDir.getFileSystem(conf);<a name="line.403"></a>
+<span class="sourceLineNo">404</span><a name="line.404"></a>
+<span class="sourceLineNo">405</span> restoreDir = new Path(restoreDir, UUID.randomUUID().toString());<a name="line.405"></a>
+<span class="sourceLineNo">406</span><a name="line.406"></a>
+<span class="sourceLineNo">407</span> // TODO: restore from record readers to parallelize.<a name="line.407"></a>
+<span class="sourceLineNo">408</span> RestoreSnapshotHelper.copySnapshotForScanner(conf, fs, rootDir, restoreDir, snapshotName);<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span> conf.set(RESTORE_DIR_KEY, restoreDir.toString());<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>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/be6f38d2/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.RecordReader.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.RecordReader.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.RecordReader.html
index 837c395..2578d72 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.RecordReader.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableSnapshotInputFormatImpl.RecordReader.html
@@ -283,137 +283,141 @@
<span class="sourceLineNo">275</span> List<HRegionInfo> regionInfos = Lists.newArrayListWithCapacity(regionManifests.size());<a name="line.275"></a>
<span class="sourceLineNo">276</span><a name="line.276"></a>
<span class="sourceLineNo">277</span> for (SnapshotRegionManifest regionManifest : regionManifests) {<a name="line.277"></a>
-<span class="sourceLineNo">278</span> regionInfos.add(HRegionInfo.convert(regionManifest.getRegionInfo()));<a name="line.278"></a>
-<span class="sourceLineNo">279</span> }<a name="line.279"></a>
-<span class="sourceLineNo">280</span> return regionInfos;<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> public static SnapshotManifest getSnapshotManifest(Configuration conf, String snapshotName,<a name="line.283"></a>
-<span class="sourceLineNo">284</span> Path rootDir, FileSystem fs) throws IOException {<a name="line.284"></a>
-<span class="sourceLineNo">285</span> Path snapshotDir = SnapshotDescriptionUtils.getCompletedSnapshotDir(snapshotName, rootDir);<a name="line.285"></a>
-<span class="sourceLineNo">286</span> SnapshotDescription snapshotDesc = SnapshotDescriptionUtils.readSnapshotInfo(fs, snapshotDir);<a name="line.286"></a>
-<span class="sourceLineNo">287</span> return SnapshotManifest.open(conf, fs, snapshotDir, snapshotDesc);<a name="line.287"></a>
-<span class="sourceLineNo">288</span> }<a name="line.288"></a>
-<span class="sourceLineNo">289</span><a name="line.289"></a>
-<span class="sourceLineNo">290</span> public static Scan extractScanFromConf(Configuration conf) throws IOException {<a name="line.290"></a>
-<span class="sourceLineNo">291</span> Scan scan = null;<a name="line.291"></a>
-<span class="sourceLineNo">292</span> if (conf.get(TableInputFormat.SCAN) != null) {<a name="line.292"></a>
-<span class="sourceLineNo">293</span> scan = TableMapReduceUtil.convertStringToScan(conf.get(TableInputFormat.SCAN));<a name="line.293"></a>
-<span class="sourceLineNo">294</span> } else if (conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST) != null) {<a name="line.294"></a>
-<span class="sourceLineNo">295</span> String[] columns =<a name="line.295"></a>
-<span class="sourceLineNo">296</span> conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST).split(" ");<a name="line.296"></a>
-<span class="sourceLineNo">297</span> scan = new Scan();<a name="line.297"></a>
-<span class="sourceLineNo">298</span> for (String col : columns) {<a name="line.298"></a>
-<span class="sourceLineNo">299</span> scan.addFamily(Bytes.toBytes(col));<a name="line.299"></a>
-<span class="sourceLineNo">300</span> }<a name="line.300"></a>
-<span class="sourceLineNo">301</span> } else {<a name="line.301"></a>
-<span class="sourceLineNo">302</span> throw new IllegalArgumentException("Unable to create scan");<a name="line.302"></a>
-<span class="sourceLineNo">303</span> }<a name="line.303"></a>
-<span class="sourceLineNo">304</span> return scan;<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> public static List<InputSplit> getSplits(Scan scan, SnapshotManifest manifest,<a name="line.307"></a>
-<span class="sourceLineNo">308</span> List<HRegionInfo> regionManifests, Path restoreDir, Configuration conf) throws IOException {<a name="line.308"></a>
-<span class="sourceLineNo">309</span> // load table descriptor<a name="line.309"></a>
-<span class="sourceLineNo">310</span> HTableDescriptor htd = manifest.getTableDescriptor();<a name="line.310"></a>
-<span class="sourceLineNo">311</span><a name="line.311"></a>
-<span class="sourceLineNo">312</span> Path tableDir = FSUtils.getTableDir(restoreDir, htd.getTableName());<a name="line.312"></a>
-<span class="sourceLineNo">313</span><a name="line.313"></a>
-<span class="sourceLineNo">314</span> List<InputSplit> splits = new ArrayList<>();<a name="line.314"></a>
-<span class="sourceLineNo">315</span> for (HRegionInfo hri : regionManifests) {<a name="line.315"></a>
-<span class="sourceLineNo">316</span> // load region descriptor<a name="line.316"></a>
+<span class="sourceLineNo">278</span> HRegionInfo hri = HRegionInfo.convert(regionManifest.getRegionInfo());<a name="line.278"></a>
+<span class="sourceLineNo">279</span> if (hri.isOffline() && (hri.isSplit() || hri.isSplitParent())) {<a name="line.279"></a>
+<span class="sourceLineNo">280</span> continue;<a name="line.280"></a>
+<span class="sourceLineNo">281</span> }<a name="line.281"></a>
+<span class="sourceLineNo">282</span> regionInfos.add(hri);<a name="line.282"></a>
+<span class="sourceLineNo">283</span> }<a name="line.283"></a>
+<span class="sourceLineNo">284</span> return regionInfos;<a name="line.284"></a>
+<span class="sourceLineNo">285</span> }<a name="line.285"></a>
+<span class="sourceLineNo">286</span><a name="line.286"></a>
+<span class="sourceLineNo">287</span> public static SnapshotManifest getSnapshotManifest(Configuration conf, String snapshotName,<a name="line.287"></a>
+<span class="sourceLineNo">288</span> Path rootDir, FileSystem fs) throws IOException {<a name="line.288"></a>
+<span class="sourceLineNo">289</span> Path snapshotDir = SnapshotDescriptionUtils.getCompletedSnapshotDir(snapshotName, rootDir);<a name="line.289"></a>
+<span class="sourceLineNo">290</span> SnapshotDescription snapshotDesc = SnapshotDescriptionUtils.readSnapshotInfo(fs, snapshotDir);<a name="line.290"></a>
+<span class="sourceLineNo">291</span> return SnapshotManifest.open(conf, fs, snapshotDir, snapshotDesc);<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> public static Scan extractScanFromConf(Configuration conf) throws IOException {<a name="line.294"></a>
+<span class="sourceLineNo">295</span> Scan scan = null;<a name="line.295"></a>
+<span class="sourceLineNo">296</span> if (conf.get(TableInputFormat.SCAN) != null) {<a name="line.296"></a>
+<span class="sourceLineNo">297</span> scan = TableMapReduceUtil.convertStringToScan(conf.get(TableInputFormat.SCAN));<a name="line.297"></a>
+<span class="sourceLineNo">298</span> } else if (conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST) != null) {<a name="line.298"></a>
+<span class="sourceLineNo">299</span> String[] columns =<a name="line.299"></a>
+<span class="sourceLineNo">300</span> conf.get(org.apache.hadoop.hbase.mapred.TableInputFormat.COLUMN_LIST).split(" ");<a name="line.300"></a>
+<span class="sourceLineNo">301</span> scan = new Scan();<a name="line.301"></a>
+<span class="sourceLineNo">302</span> for (String col : columns) {<a name="line.302"></a>
+<span class="sourceLineNo">303</span> scan.addFamily(Bytes.toBytes(col));<a name="line.303"></a>
+<span class="sourceLineNo">304</span> }<a name="line.304"></a>
+<span class="sourceLineNo">305</span> } else {<a name="line.305"></a>
+<span class="sourceLineNo">306</span> throw new IllegalArgumentException("Unable to create scan");<a name="line.306"></a>
+<span class="sourceLineNo">307</span> }<a name="line.307"></a>
+<span class="sourceLineNo">308</span> return scan;<a name="line.308"></a>
+<span class="sourceLineNo">309</span> }<a name="line.309"></a>
+<span class="sourceLineNo">310</span><a name="line.310"></a>
+<span class="sourceLineNo">311</span> public static List<InputSplit> getSplits(Scan scan, SnapshotManifest manifest,<a name="line.311"></a>
+<span class="sourceLineNo">312</span> List<HRegionInfo> regionManifests, Path restoreDir, Configuration conf) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> // load table descriptor<a name="line.313"></a>
+<span class="sourceLineNo">314</span> HTableDescriptor htd = manifest.getTableDescriptor();<a name="line.314"></a>
+<span class="sourceLineNo">315</span><a name="line.315"></a>
+<span class="sourceLineNo">316</span> Path tableDir = FSUtils.getTableDir(restoreDir, htd.getTableName());<a name="line.316"></a>
<span class="sourceLineNo">317</span><a name="line.317"></a>
-<span class="sourceLineNo">318</span> if (CellUtil.overlappingKeys(scan.getStartRow(), scan.getStopRow(), hri.getStartKey(),<a name="line.318"></a>
-<span class="sourceLineNo">319</span> hri.getEndKey())) {<a name="line.319"></a>
-<span class="sourceLineNo">320</span> // compute HDFS locations from snapshot files (which will get the locations for<a name="line.320"></a>
-<span class="sourceLineNo">321</span> // referred hfiles)<a name="line.321"></a>
-<span class="sourceLineNo">322</span> List<String> hosts = getBestLocations(conf,<a name="line.322"></a>
-<span class="sourceLineNo">323</span> HRegion.computeHDFSBlocksDistribution(conf, htd, hri, tableDir));<a name="line.323"></a>
-<span class="sourceLineNo">324</span><a name="line.324"></a>
-<span class="sourceLineNo">325</span> int len = Math.min(3, hosts.size());<a name="line.325"></a>
-<span class="sourceLineNo">326</span> hosts = hosts.subList(0, len);<a name="line.326"></a>
-<span class="sourceLineNo">327</span> splits.add(new InputSplit(htd, hri, hosts, scan, restoreDir));<a name="line.327"></a>
-<span class="sourceLineNo">328</span> }<a name="line.328"></a>
-<span class="sourceLineNo">329</span> }<a name="line.329"></a>
-<span class="sourceLineNo">330</span><a name="line.330"></a>
-<span class="sourceLineNo">331</span> return splits;<a name="line.331"></a>
-<span class="sourceLineNo">332</span><a name="line.332"></a>
-<span class="sourceLineNo">333</span> }<a name="line.333"></a>
+<span class="sourceLineNo">318</span> List<InputSplit> splits = new ArrayList<>();<a name="line.318"></a>
+<span class="sourceLineNo">319</span> for (HRegionInfo hri : regionManifests) {<a name="line.319"></a>
+<span class="sourceLineNo">320</span> // load region descriptor<a name="line.320"></a>
+<span class="sourceLineNo">321</span><a name="line.321"></a>
+<span class="sourceLineNo">322</span> if (CellUtil.overlappingKeys(scan.getStartRow(), scan.getStopRow(), hri.getStartKey(),<a name="line.322"></a>
+<span class="sourceLineNo">323</span> hri.getEndKey())) {<a name="line.323"></a>
+<span class="sourceLineNo">324</span> // compute HDFS locations from snapshot files (which will get the locations for<a name="line.324"></a>
+<span class="sourceLineNo">325</span> // referred hfiles)<a name="line.325"></a>
+<span class="sourceLineNo">326</span> List<String> hosts = getBestLocations(conf,<a name="line.326"></a>
+<span class="sourceLineNo">327</span> HRegion.computeHDFSBlocksDistribution(conf, htd, hri, tableDir));<a name="line.327"></a>
+<span class="sourceLineNo">328</span><a name="line.328"></a>
+<span class="sourceLineNo">329</span> int len = Math.min(3, hosts.size());<a name="line.329"></a>
+<span class="sourceLineNo">330</span> hosts = hosts.subList(0, len);<a name="line.330"></a>
+<span class="sourceLineNo">331</span> splits.add(new InputSplit(htd, hri, hosts, scan, restoreDir));<a name="line.331"></a>
+<span class="sourceLineNo">332</span> }<a name="line.332"></a>
+<span class="sourceLineNo">333</span> }<a name="line.333"></a>
<span class="sourceLineNo">334</span><a name="line.334"></a>
-<span class="sourceLineNo">335</span> /**<a name="line.335"></a>
-<span class="sourceLineNo">336</span> * This computes the locations to be passed from the InputSplit. MR/Yarn schedulers does not take<a name="line.336"></a>
-<span class="sourceLineNo">337</span> * weights into account, thus will treat every location passed from the input split as equal. We<a name="line.337"></a>
-<span class="sourceLineNo">338</span> * do not want to blindly pass all the locations, since we are creating one split per region, and<a name="line.338"></a>
-<span class="sourceLineNo">339</span> * the region's blocks are all distributed throughout the cluster unless favorite node assignment<a name="line.339"></a>
-<span class="sourceLineNo">340</span> * is used. On the expected stable case, only one location will contain most of the blocks as<a name="line.340"></a>
-<span class="sourceLineNo">341</span> * local.<a name="line.341"></a>
-<span class="sourceLineNo">342</span> * On the other hand, in favored node assignment, 3 nodes will contain highly local blocks. Here<a name="line.342"></a>
-<span class="sourceLineNo">343</span> * we are doing a simple heuristic, where we will pass all hosts which have at least 80%<a name="line.343"></a>
-<span class="sourceLineNo">344</span> * (hbase.tablesnapshotinputformat.locality.cutoff.multiplier) as much block locality as the top<a name="line.344"></a>
-<span class="sourceLineNo">345</span> * host with the best locality.<a name="line.345"></a>
-<span class="sourceLineNo">346</span> */<a name="line.346"></a>
-<span class="sourceLineNo">347</span> public static List<String> getBestLocations(<a name="line.347"></a>
-<span class="sourceLineNo">348</span> Configuration conf, HDFSBlocksDistribution blockDistribution) {<a name="line.348"></a>
-<span class="sourceLineNo">349</span> List<String> locations = new ArrayList<>(3);<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> HostAndWeight[] hostAndWeights = blockDistribution.getTopHostsWithWeights();<a name="line.351"></a>
-<span class="sourceLineNo">352</span><a name="line.352"></a>
-<span class="sourceLineNo">353</span> if (hostAndWeights.length == 0) {<a name="line.353"></a>
-<span class="sourceLineNo">354</span> return locations;<a name="line.354"></a>
-<span class="sourceLineNo">355</span> }<a name="line.355"></a>
+<span class="sourceLineNo">335</span> return splits;<a name="line.335"></a>
+<span class="sourceLineNo">336</span><a name="line.336"></a>
+<span class="sourceLineNo">337</span> }<a name="line.337"></a>
+<span class="sourceLineNo">338</span><a name="line.338"></a>
+<span class="sourceLineNo">339</span> /**<a name="line.339"></a>
+<span class="sourceLineNo">340</span> * This computes the locations to be passed from the InputSplit. MR/Yarn schedulers does not take<a name="line.340"></a>
+<span class="sourceLineNo">341</span> * weights into account, thus will treat every location passed from the input split as equal. We<a name="line.341"></a>
+<span class="sourceLineNo">342</span> * do not want to blindly pass all the locations, since we are creating one split per region, and<a name="line.342"></a>
+<span class="sourceLineNo">343</span> * the region's blocks are all distributed throughout the cluster unless favorite node assignment<a name="line.343"></a>
+<span class="sourceLineNo">344</span> * is used. On the expected stable case, only one location will contain most of the blocks as<a name="line.344"></a>
+<span class="sourceLineNo">345</span> * local.<a name="line.345"></a>
+<span class="sourceLineNo">346</span> * On the other hand, in favored node assignment, 3 nodes will contain highly local blocks. Here<a name="line.346"></a>
+<span class="sourceLineNo">347</span> * we are doing a simple heuristic, where we will pass all hosts which have at least 80%<a name="line.347"></a>
+<span class="sourceLineNo">348</span> * (hbase.tablesnapshotinputformat.locality.cutoff.multiplier) as much block locality as the top<a name="line.348"></a>
+<span class="sourceLineNo">349</span> * host with the best locality.<a name="line.349"></a>
+<span class="sourceLineNo">350</span> */<a name="line.350"></a>
+<span class="sourceLineNo">351</span> public static List<String> getBestLocations(<a name="line.351"></a>
+<span class="sourceLineNo">352</span> Configuration conf, HDFSBlocksDistribution blockDistribution) {<a name="line.352"></a>
+<span class="sourceLineNo">353</span> List<String> locations = new ArrayList<>(3);<a name="line.353"></a>
+<span class="sourceLineNo">354</span><a name="line.354"></a>
+<span class="sourceLineNo">355</span> HostAndWeight[] hostAndWeights = blockDistribution.getTopHostsWithWeights();<a name="line.355"></a>
<span class="sourceLineNo">356</span><a name="line.356"></a>
-<span class="sourceLineNo">357</span> HostAndWeight topHost = hostAndWeights[0];<a name="line.357"></a>
-<span class="sourceLineNo">358</span> locations.add(topHost.getHost());<a name="line.358"></a>
-<span class="sourceLineNo">359</span><a name="line.359"></a>
-<span class="sourceLineNo">360</span> // Heuristic: filter all hosts which have at least cutoffMultiplier % of block locality<a name="line.360"></a>
-<span class="sourceLineNo">361</span> double cutoffMultiplier<a name="line.361"></a>
-<span class="sourceLineNo">362</span> = conf.getFloat(LOCALITY_CUTOFF_MULTIPLIER, DEFAULT_LOCALITY_CUTOFF_MULTIPLIER);<a name="line.362"></a>
+<span class="sourceLineNo">357</span> if (hostAndWeights.length == 0) {<a name="line.357"></a>
+<span class="sourceLineNo">358</span> return locations;<a name="line.358"></a>
+<span class="sourceLineNo">359</span> }<a name="line.359"></a>
+<span class="sourceLineNo">360</span><a name="line.360"></a>
+<span class="sourceLineNo">361</span> HostAndWeight topHost = hostAndWeights[0];<a name="line.361"></a>
+<span class="sourceLineNo">362</span> locations.add(topHost.getHost());<a name="line.362"></a>
<span class="sourceLineNo">363</span><a name="line.363"></a>
-<span class="sourceLineNo">364</span> double filterWeight = topHost.getWeight() * cutoffMultiplier;<a name="line.364"></a>
-<span class="sourceLineNo">365</span><a name="line.365"></a>
-<span class="sourceLineNo">366</span> for (int i = 1; i < hostAndWeights.length; i++) {<a name="line.366"></a>
-<span class="sourceLineNo">367</span> if (hostAndWeights[i].getWeight() >= filterWeight) {<a name="line.367"></a>
-<span class="sourceLineNo">368</span> locations.add(hostAndWeights[i].getHost());<a name="line.368"></a>
-<span class="sourceLineNo">369</span> } else {<a name="line.369"></a>
-<span class="sourceLineNo">370</span> break;<a name="line.370"></a>
-<span class="sourceLineNo">371</span> }<a name="line.371"></a>
-<span class="sourceLineNo">372</span> }<a name="line.372"></a>
-<span class="sourceLineNo">373</span><a name="line.373"></a>
-<span class="sourceLineNo">374</span> return locations;<a name="line.374"></a>
-<span class="sourceLineNo">375</span> }<a name="line.375"></a>
-<span class="sourceLineNo">376</span><a name="line.376"></a>
-<span class="sourceLineNo">377</span> private static String getSnapshotName(Configuration conf) {<a name="line.377"></a>
-<span class="sourceLineNo">378</span> String snapshotName = conf.get(SNAPSHOT_NAME_KEY);<a name="line.378"></a>
-<span class="sourceLineNo">379</span> if (snapshotName == null) {<a name="line.379"></a>
-<span class="sourceLineNo">380</span> throw new IllegalArgumentException("Snapshot name must be provided");<a name="line.380"></a>
-<span class="sourceLineNo">381</span> }<a name="line.381"></a>
-<span class="sourceLineNo">382</span> return snapshotName;<a name="line.382"></a>
-<span class="sourceLineNo">383</span> }<a name="line.383"></a>
-<span class="sourceLineNo">384</span><a name="line.384"></a>
-<span class="sourceLineNo">385</span> /**<a name="line.385"></a>
-<span class="sourceLineNo">386</span> * Configures the job to use TableSnapshotInputFormat to read from a snapshot.<a name="line.386"></a>
-<span class="sourceLineNo">387</span> * @param conf the job to configuration<a name="line.387"></a>
-<span class="sourceLineNo">388</span> * @param snapshotName the name of the snapshot to read from<a name="line.388"></a>
-<span class="sourceLineNo">389</span> * @param restoreDir a temporary directory to restore the snapshot into. Current user should<a name="line.389"></a>
-<span class="sourceLineNo">390</span> * have write permissions to this directory, and this should not be a subdirectory of rootdir.<a name="line.390"></a>
-<span class="sourceLineNo">391</span> * After the job is finished, restoreDir can be deleted.<a name="line.391"></a>
-<span class="sourceLineNo">392</span> * @throws IOException if an error occurs<a name="line.392"></a>
-<span class="sourceLineNo">393</span> */<a name="line.393"></a>
-<span class="sourceLineNo">394</span> public static void setInput(Configuration conf, String snapshotName, Path restoreDir)<a name="line.394"></a>
-<span class="sourceLineNo">395</span> throws IOException {<a name="line.395"></a>
-<span class="sourceLineNo">396</span> conf.set(SNAPSHOT_NAME_KEY, snapshotName);<a name="line.396"></a>
-<span class="sourceLineNo">397</span><a name="line.397"></a>
-<span class="sourceLineNo">398</span> Path rootDir = FSUtils.getRootDir(conf);<a name="line.398"></a>
-<span class="sourceLineNo">399</span> FileSystem fs = rootDir.getFileSystem(conf);<a name="line.399"></a>
-<span class="sourceLineNo">400</span><a name="line.400"></a>
-<span class="sourceLineNo">401</span> restoreDir = new Path(restoreDir, UUID.randomUUID().toString());<a name="line.401"></a>
-<span class="sourceLineNo">402</span><a name="line.402"></a>
-<span class="sourceLineNo">403</span> // TODO: restore from record readers to parallelize.<a name="line.403"></a>
-<span class="sourceLineNo">404</span> RestoreSnapshotHelper.copySnapshotForScanner(conf, fs, rootDir, restoreDir, snapshotName);<a name="line.404"></a>
-<span class="sourceLineNo">405</span><a name="line.405"></a>
-<span class="sourceLineNo">406</span> conf.set(RESTORE_DIR_KEY, restoreDir.toString());<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">364</span> // Heuristic: filter all hosts which have at least cutoffMultiplier % of block locality<a name="line.364"></a>
+<span class="sourceLineNo">365</span> double cutoffMultiplier<a name="line.365"></a>
+<span class="sourceLineNo">366</span> = conf.getFloat(LOCALITY_CUTOFF_MULTIPLIER, DEFAULT_LOCALITY_CUTOFF_MULTIPLIER);<a name="line.366"></a>
+<span class="sourceLineNo">367</span><a name="line.367"></a>
+<span class="sourceLineNo">368</span> double filterWeight = topHost.getWeight() * cutoffMultiplier;<a name="line.368"></a>
+<span class="sourceLineNo">369</span><a name="line.369"></a>
+<span class="sourceLineNo">370</span> for (int i = 1; i < hostAndWeights.length; i++) {<a name="line.370"></a>
+<span class="sourceLineNo">371</span> if (hostAndWeights[i].getWeight() >= filterWeight) {<a name="line.371"></a>
+<span class="sourceLineNo">372</span> locations.add(hostAndWeights[i].getHost());<a name="line.372"></a>
+<span class="sourceLineNo">373</span> } else {<a name="line.373"></a>
+<span class="sourceLineNo">374</span> break;<a name="line.374"></a>
+<span class="sourceLineNo">375</span> }<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> return locations;<a name="line.378"></a>
+<span class="sourceLineNo">379</span> }<a name="line.379"></a>
+<span class="sourceLineNo">380</span><a name="line.380"></a>
+<span class="sourceLineNo">381</span> private static String getSnapshotName(Configuration conf) {<a name="line.381"></a>
+<span class="sourceLineNo">382</span> String snapshotName = conf.get(SNAPSHOT_NAME_KEY);<a name="line.382"></a>
+<span class="sourceLineNo">383</span> if (snapshotName == null) {<a name="line.383"></a>
+<span class="sourceLineNo">384</span> throw new IllegalArgumentException("Snapshot name must be provided");<a name="line.384"></a>
+<span class="sourceLineNo">385</span> }<a name="line.385"></a>
+<span class="sourceLineNo">386</span> return snapshotName;<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> /**<a name="line.389"></a>
+<span class="sourceLineNo">390</span> * Configures the job to use TableSnapshotInputFormat to read from a snapshot.<a name="line.390"></a>
+<span class="sourceLineNo">391</span> * @param conf the job to configuration<a name="line.391"></a>
+<span class="sourceLineNo">392</span> * @param snapshotName the name of the snapshot to read from<a name="line.392"></a>
+<span class="sourceLineNo">393</span> * @param restoreDir a temporary directory to restore the snapshot into. Current user should<a name="line.393"></a>
+<span class="sourceLineNo">394</span> * have write permissions to this directory, and this should not be a subdirectory of rootdir.<a name="line.394"></a>
+<span class="sourceLineNo">395</span> * After the job is finished, restoreDir can be deleted.<a name="line.395"></a>
+<span class="sourceLineNo">396</span> * @throws IOException if an error occurs<a name="line.396"></a>
+<span class="sourceLineNo">397</span> */<a name="line.397"></a>
+<span class="sourceLineNo">398</span> public static void setInput(Configuration conf, String snapshotName, Path restoreDir)<a name="line.398"></a>
+<span class="sourceLineNo">399</span> throws IOException {<a name="line.399"></a>
+<span class="sourceLineNo">400</span> conf.set(SNAPSHOT_NAME_KEY, snapshotName);<a name="line.400"></a>
+<span class="sourceLineNo">401</span><a name="line.401"></a>
+<span class="sourceLineNo">402</span> Path rootDir = FSUtils.getRootDir(conf);<a name="line.402"></a>
+<span class="sourceLineNo">403</span> FileSystem fs = rootDir.getFileSystem(conf);<a name="line.403"></a>
+<span class="sourceLineNo">404</span><a name="line.404"></a>
+<span class="sourceLineNo">405</span> restoreDir = new Path(restoreDir, UUID.randomUUID().toString());<a name="line.405"></a>
+<span class="sourceLineNo">406</span><a name="line.406"></a>
+<span class="sourceLineNo">407</span> // TODO: restore from record readers to parallelize.<a name="line.407"></a>
+<span class="sourceLineNo">408</span> RestoreSnapshotHelper.copySnapshotForScanner(conf, fs, rootDir, restoreDir, snapshotName);<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span> conf.set(RESTORE_DIR_KEY, restoreDir.toString());<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>