http://git-wip-us.apache.org/repos/asf/hbase/blob/fc29f802/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HLogInputFormat.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HLogInputFormat.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HLogInputFormat.html index 90d9303..fceace6 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HLogInputFormat.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HLogInputFormat.html @@ -42,7 +42,7 @@ <span class="sourceLineNo">034</span>/**<a name="line.34"></a> <span class="sourceLineNo">035</span> * Simple {@link InputFormat} for {@link org.apache.hadoop.hbase.wal.WAL} <a name="line.35"></a> <span class="sourceLineNo">036</span> * files.<a name="line.36"></a> -<span class="sourceLineNo">037</span> * @deprecated use {@link WALInputFormat}<a name="line.37"></a> +<span class="sourceLineNo">037</span> * @deprecated use {@link WALInputFormat}. Remove in hadoop 3.0<a name="line.37"></a> <span class="sourceLineNo">038</span> */<a name="line.38"></a> <span class="sourceLineNo">039</span>@Deprecated<a name="line.39"></a> <span class="sourceLineNo">040</span>@InterfaceAudience.Public<a name="line.40"></a>
http://git-wip-us.apache.org/repos/asf/hbase/blob/fc29f802/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HRegionPartitioner.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HRegionPartitioner.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HRegionPartitioner.html index ce2449b..8c0c32b 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HRegionPartitioner.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HRegionPartitioner.html @@ -52,7 +52,7 @@ <span class="sourceLineNo">044</span> * <p>This class is not suitable as partitioner creating hfiles<a name="line.44"></a> <span class="sourceLineNo">045</span> * for incremental bulk loads as region spread will likely change between time of<a name="line.45"></a> <span class="sourceLineNo">046</span> * hfile creation and load time. See {@link LoadIncrementalHFiles}<a name="line.46"></a> -<span class="sourceLineNo">047</span> * and <a href="http://hbase.apache.org/docs/current/bulk-loads.html">Bulk Load</a>.<a name="line.47"></a> +<span class="sourceLineNo">047</span> * and <a href="http://hbase.apache.org/book.html#arch.bulk.load">Bulk Load</a>.<a name="line.47"></a> <span class="sourceLineNo">048</span> *<a name="line.48"></a> <span class="sourceLineNo">049</span> * @param <KEY> The type of the key.<a name="line.49"></a> <span class="sourceLineNo">050</span> * @param <VALUE> The type of the value.<a name="line.50"></a> http://git-wip-us.apache.org/repos/asf/hbase/blob/fc29f802/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/KeyValueSortReducer.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/KeyValueSortReducer.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/KeyValueSortReducer.html index 2181634..807f600 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/KeyValueSortReducer.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/KeyValueSortReducer.html @@ -40,7 +40,7 @@ <span class="sourceLineNo">032</span> * Reads in all KeyValues from passed Iterator, sorts them, then emits<a name="line.32"></a> <span class="sourceLineNo">033</span> * KeyValues in sorted order. If lots of columns per row, it will use lots of<a name="line.33"></a> <span class="sourceLineNo">034</span> * memory sorting.<a name="line.34"></a> -<span class="sourceLineNo">035</span> * @see HFileOutputFormat<a name="line.35"></a> +<span class="sourceLineNo">035</span> * @see HFileOutputFormat2<a name="line.35"></a> <span class="sourceLineNo">036</span> */<a name="line.36"></a> <span class="sourceLineNo">037</span>@InterfaceAudience.Public<a name="line.37"></a> <span class="sourceLineNo">038</span>@InterfaceStability.Stable<a name="line.38"></a> http://git-wip-us.apache.org/repos/asf/hbase/blob/fc29f802/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html index 1758568..305bf89 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/PutSortReducer.html @@ -46,7 +46,7 @@ <span class="sourceLineNo">038</span> * Reads in all Puts from passed Iterator, sorts them, then emits<a name="line.38"></a> <span class="sourceLineNo">039</span> * Puts in sorted order. If lots of columns per row, it will use lots of<a name="line.39"></a> <span class="sourceLineNo">040</span> * memory sorting.<a name="line.40"></a> -<span class="sourceLineNo">041</span> * @see HFileOutputFormat<a name="line.41"></a> +<span class="sourceLineNo">041</span> * @see HFileOutputFormat2<a name="line.41"></a> <span class="sourceLineNo">042</span> * @see KeyValueSortReducer<a name="line.42"></a> <span class="sourceLineNo">043</span> */<a name="line.43"></a> <span class="sourceLineNo">044</span>@InterfaceAudience.Public<a name="line.44"></a> http://git-wip-us.apache.org/repos/asf/hbase/blob/fc29f802/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.html index d070030..5344a96 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.html @@ -352,363 +352,331 @@ <span class="sourceLineNo">344</span> }<a name="line.344"></a> <span class="sourceLineNo">345</span> }<a name="line.345"></a> <span class="sourceLineNo">346</span><a name="line.346"></a> -<span class="sourceLineNo">347</span> /**<a name="line.347"></a> -<span class="sourceLineNo">348</span> * @deprecated mistakenly made public in 0.98.7. scope will change to package-private<a name="line.348"></a> -<span class="sourceLineNo">349</span> */<a name="line.349"></a> -<span class="sourceLineNo">350</span> @Deprecated<a name="line.350"></a> -<span class="sourceLineNo">351</span> public String reverseDNS(InetAddress ipAddress) throws NamingException, UnknownHostException {<a name="line.351"></a> -<span class="sourceLineNo">352</span> String hostName = this.reverseDNSCacheMap.get(ipAddress);<a name="line.352"></a> -<span class="sourceLineNo">353</span> if (hostName == null) {<a name="line.353"></a> -<span class="sourceLineNo">354</span> String ipAddressString = null;<a name="line.354"></a> -<span class="sourceLineNo">355</span> try {<a name="line.355"></a> -<span class="sourceLineNo">356</span> ipAddressString = DNS.reverseDns(ipAddress, null);<a name="line.356"></a> -<span class="sourceLineNo">357</span> } catch (Exception e) {<a name="line.357"></a> -<span class="sourceLineNo">358</span> // We can use InetAddress in case the jndi failed to pull up the reverse DNS entry from the<a name="line.358"></a> -<span class="sourceLineNo">359</span> // name service. Also, in case of ipv6, we need to use the InetAddress since resolving<a name="line.359"></a> -<span class="sourceLineNo">360</span> // reverse DNS using jndi doesn't work well with ipv6 addresses.<a name="line.360"></a> -<span class="sourceLineNo">361</span> ipAddressString = InetAddress.getByName(ipAddress.getHostAddress()).getHostName();<a name="line.361"></a> -<span class="sourceLineNo">362</span> }<a name="line.362"></a> -<span class="sourceLineNo">363</span> if (ipAddressString == null) throw new UnknownHostException("No host found for " + ipAddress);<a name="line.363"></a> -<span class="sourceLineNo">364</span> hostName = Strings.domainNamePointerToHostName(ipAddressString);<a name="line.364"></a> -<span class="sourceLineNo">365</span> this.reverseDNSCacheMap.put(ipAddress, hostName);<a name="line.365"></a> -<span class="sourceLineNo">366</span> }<a name="line.366"></a> -<span class="sourceLineNo">367</span> return hostName;<a name="line.367"></a> -<span class="sourceLineNo">368</span> }<a name="line.368"></a> -<span class="sourceLineNo">369</span><a name="line.369"></a> -<span class="sourceLineNo">370</span> /**<a name="line.370"></a> -<span class="sourceLineNo">371</span> * Calculates the number of MapReduce input splits for the map tasks. The number of<a name="line.371"></a> -<span class="sourceLineNo">372</span> * MapReduce input splits depends on the average region size and the "data skew ratio" user set in<a name="line.372"></a> -<span class="sourceLineNo">373</span> * configuration.<a name="line.373"></a> -<span class="sourceLineNo">374</span> *<a name="line.374"></a> -<span class="sourceLineNo">375</span> * @param list The list of input splits before balance.<a name="line.375"></a> -<span class="sourceLineNo">376</span> * @param context The current job context.<a name="line.376"></a> -<span class="sourceLineNo">377</span> * @param average The average size of all regions .<a name="line.377"></a> -<span class="sourceLineNo">378</span> * @return The list of input splits.<a name="line.378"></a> -<span class="sourceLineNo">379</span> * @throws IOException When creating the list of splits fails.<a name="line.379"></a> -<span class="sourceLineNo">380</span> * @see org.apache.hadoop.mapreduce.InputFormat#getSplits(<a name="line.380"></a> -<span class="sourceLineNo">381</span> * org.apache.hadoop.mapreduce.JobContext)<a name="line.381"></a> -<span class="sourceLineNo">382</span> */<a name="line.382"></a> -<span class="sourceLineNo">383</span> private List<InputSplit> calculateRebalancedSplits(List<InputSplit> list, JobContext context,<a name="line.383"></a> -<span class="sourceLineNo">384</span> long average) throws IOException {<a name="line.384"></a> -<span class="sourceLineNo">385</span> List<InputSplit> resultList = new ArrayList<InputSplit>();<a name="line.385"></a> -<span class="sourceLineNo">386</span> Configuration conf = context.getConfiguration();<a name="line.386"></a> -<span class="sourceLineNo">387</span> //The default data skew ratio is 3<a name="line.387"></a> -<span class="sourceLineNo">388</span> long dataSkewRatio = conf.getLong(INPUT_AUTOBALANCE_MAXSKEWRATIO, 3);<a name="line.388"></a> -<span class="sourceLineNo">389</span> //It determines which mode to use: text key mode or binary key mode. The default is text mode.<a name="line.389"></a> -<span class="sourceLineNo">390</span> boolean isTextKey = context.getConfiguration().getBoolean(TABLE_ROW_TEXTKEY, true);<a name="line.390"></a> -<span class="sourceLineNo">391</span> long dataSkewThreshold = dataSkewRatio * average;<a name="line.391"></a> -<span class="sourceLineNo">392</span> int count = 0;<a name="line.392"></a> -<span class="sourceLineNo">393</span> while (count < list.size()) {<a name="line.393"></a> -<span class="sourceLineNo">394</span> TableSplit ts = (TableSplit)list.get(count);<a name="line.394"></a> -<span class="sourceLineNo">395</span> TableName tableName = ts.getTable();<a name="line.395"></a> -<span class="sourceLineNo">396</span> String regionLocation = ts.getRegionLocation();<a name="line.396"></a> -<span class="sourceLineNo">397</span> long regionSize = ts.getLength();<a name="line.397"></a> -<span class="sourceLineNo">398</span> if (regionSize >= dataSkewThreshold) {<a name="line.398"></a> -<span class="sourceLineNo">399</span> // if the current region size is large than the data skew threshold,<a name="line.399"></a> -<span class="sourceLineNo">400</span> // split the region into two MapReduce input splits.<a name="line.400"></a> -<span class="sourceLineNo">401</span> byte[] splitKey = getSplitKey(ts.getStartRow(), ts.getEndRow(), isTextKey);<a name="line.401"></a> -<span class="sourceLineNo">402</span> //Set the size of child TableSplit as 1/2 of the region size. The exact size of the<a name="line.402"></a> -<span class="sourceLineNo">403</span> // MapReduce input splits is not far off.<a name="line.403"></a> -<span class="sourceLineNo">404</span> TableSplit t1 = new TableSplit(tableName, ts.getStartRow(), splitKey, regionLocation,<a name="line.404"></a> -<span class="sourceLineNo">405</span> regionSize / 2);<a name="line.405"></a> -<span class="sourceLineNo">406</span> TableSplit t2 = new TableSplit(tableName, splitKey, ts.getEndRow(), regionLocation,<a name="line.406"></a> -<span class="sourceLineNo">407</span> regionSize - regionSize / 2);<a name="line.407"></a> -<span class="sourceLineNo">408</span> resultList.add(t1);<a name="line.408"></a> -<span class="sourceLineNo">409</span> resultList.add(t2);<a name="line.409"></a> -<span class="sourceLineNo">410</span> count++;<a name="line.410"></a> -<span class="sourceLineNo">411</span> } else if (regionSize >= average) {<a name="line.411"></a> -<span class="sourceLineNo">412</span> // if the region size between average size and data skew threshold size,<a name="line.412"></a> -<span class="sourceLineNo">413</span> // make this region as one MapReduce input split.<a name="line.413"></a> -<span class="sourceLineNo">414</span> resultList.add(ts);<a name="line.414"></a> -<span class="sourceLineNo">415</span> count++;<a name="line.415"></a> -<span class="sourceLineNo">416</span> } else {<a name="line.416"></a> -<span class="sourceLineNo">417</span> // if the total size of several small continuous regions less than the average region size,<a name="line.417"></a> -<span class="sourceLineNo">418</span> // combine them into one MapReduce input split.<a name="line.418"></a> -<span class="sourceLineNo">419</span> long totalSize = regionSize;<a name="line.419"></a> -<span class="sourceLineNo">420</span> byte[] splitStartKey = ts.getStartRow();<a name="line.420"></a> -<span class="sourceLineNo">421</span> byte[] splitEndKey = ts.getEndRow();<a name="line.421"></a> -<span class="sourceLineNo">422</span> count++;<a name="line.422"></a> -<span class="sourceLineNo">423</span> for (; count < list.size(); count++) {<a name="line.423"></a> -<span class="sourceLineNo">424</span> TableSplit nextRegion = (TableSplit)list.get(count);<a name="line.424"></a> -<span class="sourceLineNo">425</span> long nextRegionSize = nextRegion.getLength();<a name="line.425"></a> -<span class="sourceLineNo">426</span> if (totalSize + nextRegionSize <= dataSkewThreshold) {<a name="line.426"></a> -<span class="sourceLineNo">427</span> totalSize = totalSize + nextRegionSize;<a name="line.427"></a> -<span class="sourceLineNo">428</span> splitEndKey = nextRegion.getEndRow();<a name="line.428"></a> -<span class="sourceLineNo">429</span> } else {<a name="line.429"></a> -<span class="sourceLineNo">430</span> break;<a name="line.430"></a> -<span class="sourceLineNo">431</span> }<a name="line.431"></a> -<span class="sourceLineNo">432</span> }<a name="line.432"></a> -<span class="sourceLineNo">433</span> TableSplit t = new TableSplit(tableName, splitStartKey, splitEndKey,<a name="line.433"></a> -<span class="sourceLineNo">434</span> regionLocation, totalSize);<a name="line.434"></a> -<span class="sourceLineNo">435</span> resultList.add(t);<a name="line.435"></a> -<span class="sourceLineNo">436</span> }<a name="line.436"></a> -<span class="sourceLineNo">437</span> }<a name="line.437"></a> -<span class="sourceLineNo">438</span> return resultList;<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> * select a split point in the region. The selection of the split point is based on an uniform<a name="line.442"></a> -<span class="sourceLineNo">443</span> * distribution assumption for the keys in a region.<a name="line.443"></a> -<span class="sourceLineNo">444</span> * Here are some examples:<a name="line.444"></a> -<span class="sourceLineNo">445</span> * startKey: aaabcdefg endKey: aaafff split point: aaad<a name="line.445"></a> -<span class="sourceLineNo">446</span> * startKey: 111000 endKey: 1125790 split point: 111b<a name="line.446"></a> -<span class="sourceLineNo">447</span> * startKey: 1110 endKey: 1120 split point: 111_<a name="line.447"></a> -<span class="sourceLineNo">448</span> * startKey: binary key { 13, -19, 126, 127 }, endKey: binary key { 13, -19, 127, 0 },<a name="line.448"></a> -<span class="sourceLineNo">449</span> * split point: binary key { 13, -19, 127, -64 }<a name="line.449"></a> -<span class="sourceLineNo">450</span> * Set this function as "public static", make it easier for test.<a name="line.450"></a> -<span class="sourceLineNo">451</span> *<a name="line.451"></a> -<span class="sourceLineNo">452</span> * @param start Start key of the region<a name="line.452"></a> -<span class="sourceLineNo">453</span> * @param end End key of the region<a name="line.453"></a> -<span class="sourceLineNo">454</span> * @param isText It determines to use text key mode or binary key mode<a name="line.454"></a> -<span class="sourceLineNo">455</span> * @return The split point in the region.<a name="line.455"></a> -<span class="sourceLineNo">456</span> */<a name="line.456"></a> -<span class="sourceLineNo">457</span> @InterfaceAudience.Private<a name="line.457"></a> -<span class="sourceLineNo">458</span> public static byte[] getSplitKey(byte[] start, byte[] end, boolean isText) {<a name="line.458"></a> -<span class="sourceLineNo">459</span> byte upperLimitByte;<a name="line.459"></a> -<span class="sourceLineNo">460</span> byte lowerLimitByte;<a name="line.460"></a> -<span class="sourceLineNo">461</span> //Use text mode or binary mode.<a name="line.461"></a> -<span class="sourceLineNo">462</span> if (isText) {<a name="line.462"></a> -<span class="sourceLineNo">463</span> //The range of text char set in ASCII is [32,126], the lower limit is space and the upper<a name="line.463"></a> -<span class="sourceLineNo">464</span> // limit is '~'.<a name="line.464"></a> -<span class="sourceLineNo">465</span> upperLimitByte = '~';<a name="line.465"></a> -<span class="sourceLineNo">466</span> lowerLimitByte = ' ';<a name="line.466"></a> -<span class="sourceLineNo">467</span> } else {<a name="line.467"></a> -<span class="sourceLineNo">468</span> upperLimitByte = Byte.MAX_VALUE;<a name="line.468"></a> -<span class="sourceLineNo">469</span> lowerLimitByte = Byte.MIN_VALUE;<a name="line.469"></a> -<span class="sourceLineNo">470</span> }<a name="line.470"></a> -<span class="sourceLineNo">471</span> // For special case<a name="line.471"></a> -<span class="sourceLineNo">472</span> // Example 1 : startkey=null, endkey="hhhqqqwww", splitKey="h"<a name="line.472"></a> -<span class="sourceLineNo">473</span> // Example 2 (text key mode): startKey="ffffaaa", endKey=null, splitkey="f~~~~~~"<a name="line.473"></a> -<span class="sourceLineNo">474</span> if (start.length == 0 && end.length == 0){<a name="line.474"></a> -<span class="sourceLineNo">475</span> return new byte[]{(byte) ((lowerLimitByte + upperLimitByte) / 2)};<a name="line.475"></a> -<span class="sourceLineNo">476</span> }<a name="line.476"></a> -<span class="sourceLineNo">477</span> if (start.length == 0 && end.length != 0){<a name="line.477"></a> -<span class="sourceLineNo">478</span> return new byte[]{ end[0] };<a name="line.478"></a> -<span class="sourceLineNo">479</span> }<a name="line.479"></a> -<span class="sourceLineNo">480</span> if (start.length != 0 && end.length == 0){<a name="line.480"></a> -<span class="sourceLineNo">481</span> byte[] result =new byte[start.length];<a name="line.481"></a> -<span class="sourceLineNo">482</span> result[0]=start[0];<a name="line.482"></a> -<span class="sourceLineNo">483</span> for (int k = 1; k < start.length; k++){<a name="line.483"></a> -<span class="sourceLineNo">484</span> result[k] = upperLimitByte;<a name="line.484"></a> -<span class="sourceLineNo">485</span> }<a name="line.485"></a> -<span class="sourceLineNo">486</span> return result;<a name="line.486"></a> -<span class="sourceLineNo">487</span> }<a name="line.487"></a> -<span class="sourceLineNo">488</span> // A list to store bytes in split key<a name="line.488"></a> -<span class="sourceLineNo">489</span> List resultBytesList = new ArrayList();<a name="line.489"></a> -<span class="sourceLineNo">490</span> int maxLength = start.length > end.length ? start.length : end.length;<a name="line.490"></a> -<span class="sourceLineNo">491</span> for (int i = 0; i < maxLength; i++) {<a name="line.491"></a> -<span class="sourceLineNo">492</span> //calculate the midpoint byte between the first difference<a name="line.492"></a> -<span class="sourceLineNo">493</span> //for example: "11ae" and "11chw", the midpoint is "11b"<a name="line.493"></a> -<span class="sourceLineNo">494</span> //another example: "11ae" and "11bhw", the first different byte is 'a' and 'b',<a name="line.494"></a> -<span class="sourceLineNo">495</span> // there is no midpoint between 'a' and 'b', so we need to check the next byte.<a name="line.495"></a> -<span class="sourceLineNo">496</span> if (start[i] == end[i]) {<a name="line.496"></a> -<span class="sourceLineNo">497</span> resultBytesList.add(start[i]);<a name="line.497"></a> -<span class="sourceLineNo">498</span> //For special case like: startKey="aaa", endKey="aaaz", splitKey="aaaM"<a name="line.498"></a> -<span class="sourceLineNo">499</span> if (i + 1 == start.length) {<a name="line.499"></a> -<span class="sourceLineNo">500</span> resultBytesList.add((byte) ((lowerLimitByte + end[i + 1]) / 2));<a name="line.500"></a> -<span class="sourceLineNo">501</span> break;<a name="line.501"></a> -<span class="sourceLineNo">502</span> }<a name="line.502"></a> -<span class="sourceLineNo">503</span> } else {<a name="line.503"></a> -<span class="sourceLineNo">504</span> //if the two bytes differ by 1, like ['a','b'], We need to check the next byte to find<a name="line.504"></a> -<span class="sourceLineNo">505</span> // the midpoint.<a name="line.505"></a> -<span class="sourceLineNo">506</span> if ((int)end[i] - (int)start[i] == 1) {<a name="line.506"></a> -<span class="sourceLineNo">507</span> //get next byte after the first difference<a name="line.507"></a> -<span class="sourceLineNo">508</span> byte startNextByte = (i + 1 < start.length) ? start[i + 1] : lowerLimitByte;<a name="line.508"></a> -<span class="sourceLineNo">509</span> byte endNextByte = (i + 1 < end.length) ? end[i + 1] : lowerLimitByte;<a name="line.509"></a> -<span class="sourceLineNo">510</span> int byteRange = (upperLimitByte - startNextByte) + (endNextByte - lowerLimitByte) + 1;<a name="line.510"></a> -<span class="sourceLineNo">511</span> int halfRange = byteRange / 2;<a name="line.511"></a> -<span class="sourceLineNo">512</span> if ((int)startNextByte + halfRange > (int)upperLimitByte) {<a name="line.512"></a> -<span class="sourceLineNo">513</span> resultBytesList.add(end[i]);<a name="line.513"></a> -<span class="sourceLineNo">514</span> resultBytesList.add((byte) (startNextByte + halfRange - upperLimitByte +<a name="line.514"></a> -<span class="sourceLineNo">515</span> lowerLimitByte));<a name="line.515"></a> -<span class="sourceLineNo">516</span> } else {<a name="line.516"></a> -<span class="sourceLineNo">517</span> resultBytesList.add(start[i]);<a name="line.517"></a> -<span class="sourceLineNo">518</span> resultBytesList.add((byte) (startNextByte + halfRange));<a name="line.518"></a> -<span class="sourceLineNo">519</span> }<a name="line.519"></a> -<span class="sourceLineNo">520</span> } else {<a name="line.520"></a> -<span class="sourceLineNo">521</span> //calculate the midpoint key by the fist different byte (normal case),<a name="line.521"></a> -<span class="sourceLineNo">522</span> // like "11ae" and "11chw", the midpoint is "11b"<a name="line.522"></a> -<span class="sourceLineNo">523</span> resultBytesList.add((byte) ((start[i] + end[i]) / 2));<a name="line.523"></a> -<span class="sourceLineNo">524</span> }<a name="line.524"></a> -<span class="sourceLineNo">525</span> break;<a name="line.525"></a> -<span class="sourceLineNo">526</span> }<a name="line.526"></a> -<span class="sourceLineNo">527</span> }<a name="line.527"></a> -<span class="sourceLineNo">528</span> //transform the List of bytes to byte[]<a name="line.528"></a> -<span class="sourceLineNo">529</span> byte[] result = new byte[resultBytesList.size()];<a name="line.529"></a> -<span class="sourceLineNo">530</span> for (int k = 0; k < resultBytesList.size(); k++) {<a name="line.530"></a> -<span class="sourceLineNo">531</span> result[k] = (byte) resultBytesList.get(k);<a name="line.531"></a> -<span class="sourceLineNo">532</span> }<a name="line.532"></a> -<span class="sourceLineNo">533</span> return result;<a name="line.533"></a> -<span class="sourceLineNo">534</span> }<a name="line.534"></a> -<span class="sourceLineNo">535</span><a name="line.535"></a> -<span class="sourceLineNo">536</span> /**<a name="line.536"></a> -<span class="sourceLineNo">537</span> * Test if the given region is to be included in the InputSplit while splitting<a name="line.537"></a> -<span class="sourceLineNo">538</span> * the regions of a table.<a name="line.538"></a> -<span class="sourceLineNo">539</span> * <p><a name="line.539"></a> -<span class="sourceLineNo">540</span> * This optimization is effective when there is a specific reasoning to exclude an entire region from the M-R job,<a name="line.540"></a> -<span class="sourceLineNo">541</span> * (and hence, not contributing to the InputSplit), given the start and end keys of the same. <br><a name="line.541"></a> -<span class="sourceLineNo">542</span> * Useful when we need to remember the last-processed top record and revisit the [last, current) interval for M-R processing,<a name="line.542"></a> -<span class="sourceLineNo">543</span> * continuously. In addition to reducing InputSplits, reduces the load on the region server as well, due to the ordering of the keys.<a name="line.543"></a> -<span class="sourceLineNo">544</span> * <br><a name="line.544"></a> -<span class="sourceLineNo">545</span> * <br><a name="line.545"></a> -<span class="sourceLineNo">546</span> * Note: It is possible that <code>endKey.length() == 0 </code> , for the last (recent) region.<a name="line.546"></a> -<span class="sourceLineNo">547</span> * <br><a name="line.547"></a> -<span class="sourceLineNo">548</span> * Override this method, if you want to bulk exclude regions altogether from M-R. By default, no region is excluded( i.e. all regions are included).<a name="line.548"></a> -<span class="sourceLineNo">549</span> *<a name="line.549"></a> +<span class="sourceLineNo">347</span> String reverseDNS(InetAddress ipAddress) throws NamingException, UnknownHostException {<a name="line.347"></a> +<span class="sourceLineNo">348</span> String hostName = this.reverseDNSCacheMap.get(ipAddress);<a name="line.348"></a> +<span class="sourceLineNo">349</span> if (hostName == null) {<a name="line.349"></a> +<span class="sourceLineNo">350</span> String ipAddressString = null;<a name="line.350"></a> +<span class="sourceLineNo">351</span> try {<a name="line.351"></a> +<span class="sourceLineNo">352</span> ipAddressString = DNS.reverseDns(ipAddress, null);<a name="line.352"></a> +<span class="sourceLineNo">353</span> } catch (Exception e) {<a name="line.353"></a> +<span class="sourceLineNo">354</span> // We can use InetAddress in case the jndi failed to pull up the reverse DNS entry from the<a name="line.354"></a> +<span class="sourceLineNo">355</span> // name service. Also, in case of ipv6, we need to use the InetAddress since resolving<a name="line.355"></a> +<span class="sourceLineNo">356</span> // reverse DNS using jndi doesn't work well with ipv6 addresses.<a name="line.356"></a> +<span class="sourceLineNo">357</span> ipAddressString = InetAddress.getByName(ipAddress.getHostAddress()).getHostName();<a name="line.357"></a> +<span class="sourceLineNo">358</span> }<a name="line.358"></a> +<span class="sourceLineNo">359</span> if (ipAddressString == null) throw new UnknownHostException("No host found for " + ipAddress);<a name="line.359"></a> +<span class="sourceLineNo">360</span> hostName = Strings.domainNamePointerToHostName(ipAddressString);<a name="line.360"></a> +<span class="sourceLineNo">361</span> this.reverseDNSCacheMap.put(ipAddress, hostName);<a name="line.361"></a> +<span class="sourceLineNo">362</span> }<a name="line.362"></a> +<span class="sourceLineNo">363</span> return hostName;<a name="line.363"></a> +<span class="sourceLineNo">364</span> }<a name="line.364"></a> +<span class="sourceLineNo">365</span><a name="line.365"></a> +<span class="sourceLineNo">366</span> /**<a name="line.366"></a> +<span class="sourceLineNo">367</span> * Calculates the number of MapReduce input splits for the map tasks. The number of<a name="line.367"></a> +<span class="sourceLineNo">368</span> * MapReduce input splits depends on the average region size and the "data skew ratio" user set in<a name="line.368"></a> +<span class="sourceLineNo">369</span> * configuration.<a name="line.369"></a> +<span class="sourceLineNo">370</span> *<a name="line.370"></a> +<span class="sourceLineNo">371</span> * @param list The list of input splits before balance.<a name="line.371"></a> +<span class="sourceLineNo">372</span> * @param context The current job context.<a name="line.372"></a> +<span class="sourceLineNo">373</span> * @param average The average size of all regions .<a name="line.373"></a> +<span class="sourceLineNo">374</span> * @return The list of input splits.<a name="line.374"></a> +<span class="sourceLineNo">375</span> * @throws IOException When creating the list of splits fails.<a name="line.375"></a> +<span class="sourceLineNo">376</span> * @see org.apache.hadoop.mapreduce.InputFormat#getSplits(<a name="line.376"></a> +<span class="sourceLineNo">377</span> * org.apache.hadoop.mapreduce.JobContext)<a name="line.377"></a> +<span class="sourceLineNo">378</span> */<a name="line.378"></a> +<span class="sourceLineNo">379</span> private List<InputSplit> calculateRebalancedSplits(List<InputSplit> list, JobContext context,<a name="line.379"></a> +<span class="sourceLineNo">380</span> long average) throws IOException {<a name="line.380"></a> +<span class="sourceLineNo">381</span> List<InputSplit> resultList = new ArrayList<InputSplit>();<a name="line.381"></a> +<span class="sourceLineNo">382</span> Configuration conf = context.getConfiguration();<a name="line.382"></a> +<span class="sourceLineNo">383</span> //The default data skew ratio is 3<a name="line.383"></a> +<span class="sourceLineNo">384</span> long dataSkewRatio = conf.getLong(INPUT_AUTOBALANCE_MAXSKEWRATIO, 3);<a name="line.384"></a> +<span class="sourceLineNo">385</span> //It determines which mode to use: text key mode or binary key mode. The default is text mode.<a name="line.385"></a> +<span class="sourceLineNo">386</span> boolean isTextKey = context.getConfiguration().getBoolean(TABLE_ROW_TEXTKEY, true);<a name="line.386"></a> +<span class="sourceLineNo">387</span> long dataSkewThreshold = dataSkewRatio * average;<a name="line.387"></a> +<span class="sourceLineNo">388</span> int count = 0;<a name="line.388"></a> +<span class="sourceLineNo">389</span> while (count < list.size()) {<a name="line.389"></a> +<span class="sourceLineNo">390</span> TableSplit ts = (TableSplit)list.get(count);<a name="line.390"></a> +<span class="sourceLineNo">391</span> TableName tableName = ts.getTable();<a name="line.391"></a> +<span class="sourceLineNo">392</span> String regionLocation = ts.getRegionLocation();<a name="line.392"></a> +<span class="sourceLineNo">393</span> long regionSize = ts.getLength();<a name="line.393"></a> +<span class="sourceLineNo">394</span> if (regionSize >= dataSkewThreshold) {<a name="line.394"></a> +<span class="sourceLineNo">395</span> // if the current region size is large than the data skew threshold,<a name="line.395"></a> +<span class="sourceLineNo">396</span> // split the region into two MapReduce input splits.<a name="line.396"></a> +<span class="sourceLineNo">397</span> byte[] splitKey = getSplitKey(ts.getStartRow(), ts.getEndRow(), isTextKey);<a name="line.397"></a> +<span class="sourceLineNo">398</span> //Set the size of child TableSplit as 1/2 of the region size. The exact size of the<a name="line.398"></a> +<span class="sourceLineNo">399</span> // MapReduce input splits is not far off.<a name="line.399"></a> +<span class="sourceLineNo">400</span> TableSplit t1 = new TableSplit(tableName, ts.getStartRow(), splitKey, regionLocation,<a name="line.400"></a> +<span class="sourceLineNo">401</span> regionSize / 2);<a name="line.401"></a> +<span class="sourceLineNo">402</span> TableSplit t2 = new TableSplit(tableName, splitKey, ts.getEndRow(), regionLocation,<a name="line.402"></a> +<span class="sourceLineNo">403</span> regionSize - regionSize / 2);<a name="line.403"></a> +<span class="sourceLineNo">404</span> resultList.add(t1);<a name="line.404"></a> +<span class="sourceLineNo">405</span> resultList.add(t2);<a name="line.405"></a> +<span class="sourceLineNo">406</span> count++;<a name="line.406"></a> +<span class="sourceLineNo">407</span> } else if (regionSize >= average) {<a name="line.407"></a> +<span class="sourceLineNo">408</span> // if the region size between average size and data skew threshold size,<a name="line.408"></a> +<span class="sourceLineNo">409</span> // make this region as one MapReduce input split.<a name="line.409"></a> +<span class="sourceLineNo">410</span> resultList.add(ts);<a name="line.410"></a> +<span class="sourceLineNo">411</span> count++;<a name="line.411"></a> +<span class="sourceLineNo">412</span> } else {<a name="line.412"></a> +<span class="sourceLineNo">413</span> // if the total size of several small continuous regions less than the average region size,<a name="line.413"></a> +<span class="sourceLineNo">414</span> // combine them into one MapReduce input split.<a name="line.414"></a> +<span class="sourceLineNo">415</span> long totalSize = regionSize;<a name="line.415"></a> +<span class="sourceLineNo">416</span> byte[] splitStartKey = ts.getStartRow();<a name="line.416"></a> +<span class="sourceLineNo">417</span> byte[] splitEndKey = ts.getEndRow();<a name="line.417"></a> +<span class="sourceLineNo">418</span> count++;<a name="line.418"></a> +<span class="sourceLineNo">419</span> for (; count < list.size(); count++) {<a name="line.419"></a> +<span class="sourceLineNo">420</span> TableSplit nextRegion = (TableSplit)list.get(count);<a name="line.420"></a> +<span class="sourceLineNo">421</span> long nextRegionSize = nextRegion.getLength();<a name="line.421"></a> +<span class="sourceLineNo">422</span> if (totalSize + nextRegionSize <= dataSkewThreshold) {<a name="line.422"></a> +<span class="sourceLineNo">423</span> totalSize = totalSize + nextRegionSize;<a name="line.423"></a> +<span class="sourceLineNo">424</span> splitEndKey = nextRegion.getEndRow();<a name="line.424"></a> +<span class="sourceLineNo">425</span> } else {<a name="line.425"></a> +<span class="sourceLineNo">426</span> break;<a name="line.426"></a> +<span class="sourceLineNo">427</span> }<a name="line.427"></a> +<span class="sourceLineNo">428</span> }<a name="line.428"></a> +<span class="sourceLineNo">429</span> TableSplit t = new TableSplit(tableName, splitStartKey, splitEndKey,<a name="line.429"></a> +<span class="sourceLineNo">430</span> regionLocation, totalSize);<a name="line.430"></a> +<span class="sourceLineNo">431</span> resultList.add(t);<a name="line.431"></a> +<span class="sourceLineNo">432</span> }<a name="line.432"></a> +<span class="sourceLineNo">433</span> }<a name="line.433"></a> +<span class="sourceLineNo">434</span> return resultList;<a name="line.434"></a> +<span class="sourceLineNo">435</span> }<a name="line.435"></a> +<span class="sourceLineNo">436</span><a name="line.436"></a> +<span class="sourceLineNo">437</span> /**<a name="line.437"></a> +<span class="sourceLineNo">438</span> * select a split point in the region. The selection of the split point is based on an uniform<a name="line.438"></a> +<span class="sourceLineNo">439</span> * distribution assumption for the keys in a region.<a name="line.439"></a> +<span class="sourceLineNo">440</span> * Here are some examples:<a name="line.440"></a> +<span class="sourceLineNo">441</span> * startKey: aaabcdefg endKey: aaafff split point: aaad<a name="line.441"></a> +<span class="sourceLineNo">442</span> * startKey: 111000 endKey: 1125790 split point: 111b<a name="line.442"></a> +<span class="sourceLineNo">443</span> * startKey: 1110 endKey: 1120 split point: 111_<a name="line.443"></a> +<span class="sourceLineNo">444</span> * startKey: binary key { 13, -19, 126, 127 }, endKey: binary key { 13, -19, 127, 0 },<a name="line.444"></a> +<span class="sourceLineNo">445</span> * split point: binary key { 13, -19, 127, -64 }<a name="line.445"></a> +<span class="sourceLineNo">446</span> * Set this function as "public static", make it easier for test.<a name="line.446"></a> +<span class="sourceLineNo">447</span> *<a name="line.447"></a> +<span class="sourceLineNo">448</span> * @param start Start key of the region<a name="line.448"></a> +<span class="sourceLineNo">449</span> * @param end End key of the region<a name="line.449"></a> +<span class="sourceLineNo">450</span> * @param isText It determines to use text key mode or binary key mode<a name="line.450"></a> +<span class="sourceLineNo">451</span> * @return The split point in the region.<a name="line.451"></a> +<span class="sourceLineNo">452</span> */<a name="line.452"></a> +<span class="sourceLineNo">453</span> @InterfaceAudience.Private<a name="line.453"></a> +<span class="sourceLineNo">454</span> public static byte[] getSplitKey(byte[] start, byte[] end, boolean isText) {<a name="line.454"></a> +<span class="sourceLineNo">455</span> byte upperLimitByte;<a name="line.455"></a> +<span class="sourceLineNo">456</span> byte lowerLimitByte;<a name="line.456"></a> +<span class="sourceLineNo">457</span> //Use text mode or binary mode.<a name="line.457"></a> +<span class="sourceLineNo">458</span> if (isText) {<a name="line.458"></a> +<span class="sourceLineNo">459</span> //The range of text char set in ASCII is [32,126], the lower limit is space and the upper<a name="line.459"></a> +<span class="sourceLineNo">460</span> // limit is '~'.<a name="line.460"></a> +<span class="sourceLineNo">461</span> upperLimitByte = '~';<a name="line.461"></a> +<span class="sourceLineNo">462</span> lowerLimitByte = ' ';<a name="line.462"></a> +<span class="sourceLineNo">463</span> } else {<a name="line.463"></a> +<span class="sourceLineNo">464</span> upperLimitByte = Byte.MAX_VALUE;<a name="line.464"></a> +<span class="sourceLineNo">465</span> lowerLimitByte = Byte.MIN_VALUE;<a name="line.465"></a> +<span class="sourceLineNo">466</span> }<a name="line.466"></a> +<span class="sourceLineNo">467</span> // For special case<a name="line.467"></a> +<span class="sourceLineNo">468</span> // Example 1 : startkey=null, endkey="hhhqqqwww", splitKey="h"<a name="line.468"></a> +<span class="sourceLineNo">469</span> // Example 2 (text key mode): startKey="ffffaaa", endKey=null, splitkey="f~~~~~~"<a name="line.469"></a> +<span class="sourceLineNo">470</span> if (start.length == 0 && end.length == 0){<a name="line.470"></a> +<span class="sourceLineNo">471</span> return new byte[]{(byte) ((lowerLimitByte + upperLimitByte) / 2)};<a name="line.471"></a> +<span class="sourceLineNo">472</span> }<a name="line.472"></a> +<span class="sourceLineNo">473</span> if (start.length == 0 && end.length != 0){<a name="line.473"></a> +<span class="sourceLineNo">474</span> return new byte[]{ end[0] };<a name="line.474"></a> +<span class="sourceLineNo">475</span> }<a name="line.475"></a> +<span class="sourceLineNo">476</span> if (start.length != 0 && end.length == 0){<a name="line.476"></a> +<span class="sourceLineNo">477</span> byte[] result =new byte[start.length];<a name="line.477"></a> +<span class="sourceLineNo">478</span> result[0]=start[0];<a name="line.478"></a> +<span class="sourceLineNo">479</span> for (int k = 1; k < start.length; k++){<a name="line.479"></a> +<span class="sourceLineNo">480</span> result[k] = upperLimitByte;<a name="line.480"></a> +<span class="sourceLineNo">481</span> }<a name="line.481"></a> +<span class="sourceLineNo">482</span> return result;<a name="line.482"></a> +<span class="sourceLineNo">483</span> }<a name="line.483"></a> +<span class="sourceLineNo">484</span> // A list to store bytes in split key<a name="line.484"></a> +<span class="sourceLineNo">485</span> List resultBytesList = new ArrayList();<a name="line.485"></a> +<span class="sourceLineNo">486</span> int maxLength = start.length > end.length ? start.length : end.length;<a name="line.486"></a> +<span class="sourceLineNo">487</span> for (int i = 0; i < maxLength; i++) {<a name="line.487"></a> +<span class="sourceLineNo">488</span> //calculate the midpoint byte between the first difference<a name="line.488"></a> +<span class="sourceLineNo">489</span> //for example: "11ae" and "11chw", the midpoint is "11b"<a name="line.489"></a> +<span class="sourceLineNo">490</span> //another example: "11ae" and "11bhw", the first different byte is 'a' and 'b',<a name="line.490"></a> +<span class="sourceLineNo">491</span> // there is no midpoint between 'a' and 'b', so we need to check the next byte.<a name="line.491"></a> +<span class="sourceLineNo">492</span> if (start[i] == end[i]) {<a name="line.492"></a> +<span class="sourceLineNo">493</span> resultBytesList.add(start[i]);<a name="line.493"></a> +<span class="sourceLineNo">494</span> //For special case like: startKey="aaa", endKey="aaaz", splitKey="aaaM"<a name="line.494"></a> +<span class="sourceLineNo">495</span> if (i + 1 == start.length) {<a name="line.495"></a> +<span class="sourceLineNo">496</span> resultBytesList.add((byte) ((lowerLimitByte + end[i + 1]) / 2));<a name="line.496"></a> +<span class="sourceLineNo">497</span> break;<a name="line.497"></a> +<span class="sourceLineNo">498</span> }<a name="line.498"></a> +<span class="sourceLineNo">499</span> } else {<a name="line.499"></a> +<span class="sourceLineNo">500</span> //if the two bytes differ by 1, like ['a','b'], We need to check the next byte to find<a name="line.500"></a> +<span class="sourceLineNo">501</span> // the midpoint.<a name="line.501"></a> +<span class="sourceLineNo">502</span> if ((int)end[i] - (int)start[i] == 1) {<a name="line.502"></a> +<span class="sourceLineNo">503</span> //get next byte after the first difference<a name="line.503"></a> +<span class="sourceLineNo">504</span> byte startNextByte = (i + 1 < start.length) ? start[i + 1] : lowerLimitByte;<a name="line.504"></a> +<span class="sourceLineNo">505</span> byte endNextByte = (i + 1 < end.length) ? end[i + 1] : lowerLimitByte;<a name="line.505"></a> +<span class="sourceLineNo">506</span> int byteRange = (upperLimitByte - startNextByte) + (endNextByte - lowerLimitByte) + 1;<a name="line.506"></a> +<span class="sourceLineNo">507</span> int halfRange = byteRange / 2;<a name="line.507"></a> +<span class="sourceLineNo">508</span> if ((int)startNextByte + halfRange > (int)upperLimitByte) {<a name="line.508"></a> +<span class="sourceLineNo">509</span> resultBytesList.add(end[i]);<a name="line.509"></a> +<span class="sourceLineNo">510</span> resultBytesList.add((byte) (startNextByte + halfRange - upperLimitByte +<a name="line.510"></a> +<span class="sourceLineNo">511</span> lowerLimitByte));<a name="line.511"></a> +<span class="sourceLineNo">512</span> } else {<a name="line.512"></a> +<span class="sourceLineNo">513</span> resultBytesList.add(start[i]);<a name="line.513"></a> +<span class="sourceLineNo">514</span> resultBytesList.add((byte) (startNextByte + halfRange));<a name="line.514"></a> +<span class="sourceLineNo">515</span> }<a name="line.515"></a> +<span class="sourceLineNo">516</span> } else {<a name="line.516"></a> +<span class="sourceLineNo">517</span> //calculate the midpoint key by the fist different byte (normal case),<a name="line.517"></a> +<span class="sourceLineNo">518</span> // like "11ae" and "11chw", the midpoint is "11b"<a name="line.518"></a> +<span class="sourceLineNo">519</span> resultBytesList.add((byte) ((start[i] + end[i]) / 2));<a name="line.519"></a> +<span class="sourceLineNo">520</span> }<a name="line.520"></a> +<span class="sourceLineNo">521</span> break;<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> //transform the List of bytes to byte[]<a name="line.524"></a> +<span class="sourceLineNo">525</span> byte[] result = new byte[resultBytesList.size()];<a name="line.525"></a> +<span class="sourceLineNo">526</span> for (int k = 0; k < resultBytesList.size(); k++) {<a name="line.526"></a> +<span class="sourceLineNo">527</span> result[k] = (byte) resultBytesList.get(k);<a name="line.527"></a> +<span class="sourceLineNo">528</span> }<a name="line.528"></a> +<span class="sourceLineNo">529</span> return result;<a name="line.529"></a> +<span class="sourceLineNo">530</span> }<a name="line.530"></a> +<span class="sourceLineNo">531</span><a name="line.531"></a> +<span class="sourceLineNo">532</span> /**<a name="line.532"></a> +<span class="sourceLineNo">533</span> * Test if the given region is to be included in the InputSplit while splitting<a name="line.533"></a> +<span class="sourceLineNo">534</span> * the regions of a table.<a name="line.534"></a> +<span class="sourceLineNo">535</span> * <p><a name="line.535"></a> +<span class="sourceLineNo">536</span> * This optimization is effective when there is a specific reasoning to exclude an entire region from the M-R job,<a name="line.536"></a> +<span class="sourceLineNo">537</span> * (and hence, not contributing to the InputSplit), given the start and end keys of the same. <br><a name="line.537"></a> +<span class="sourceLineNo">538</span> * Useful when we need to remember the last-processed top record and revisit the [last, current) interval for M-R processing,<a name="line.538"></a> +<span class="sourceLineNo">539</span> * continuously. In addition to reducing InputSplits, reduces the load on the region server as well, due to the ordering of the keys.<a name="line.539"></a> +<span class="sourceLineNo">540</span> * <br><a name="line.540"></a> +<span class="sourceLineNo">541</span> * <br><a name="line.541"></a> +<span class="sourceLineNo">542</span> * Note: It is possible that <code>endKey.length() == 0 </code> , for the last (recent) region.<a name="line.542"></a> +<span class="sourceLineNo">543</span> * <br><a name="line.543"></a> +<span class="sourceLineNo">544</span> * Override this method, if you want to bulk exclude regions altogether from M-R. By default, no region is excluded( i.e. all regions are included).<a name="line.544"></a> +<span class="sourceLineNo">545</span> *<a name="line.545"></a> +<span class="sourceLineNo">546</span> *<a name="line.546"></a> +<span class="sourceLineNo">547</span> * @param startKey Start key of the region<a name="line.547"></a> +<span class="sourceLineNo">548</span> * @param endKey End key of the region<a name="line.548"></a> +<span class="sourceLineNo">549</span> * @return true, if this region needs to be included as part of the input (default).<a name="line.549"></a> <span class="sourceLineNo">550</span> *<a name="line.550"></a> -<span class="sourceLineNo">551</span> * @param startKey Start key of the region<a name="line.551"></a> -<span class="sourceLineNo">552</span> * @param endKey End key of the region<a name="line.552"></a> -<span class="sourceLineNo">553</span> * @return true, if this region needs to be included as part of the input (default).<a name="line.553"></a> -<span class="sourceLineNo">554</span> *<a name="line.554"></a> -<span class="sourceLineNo">555</span> */<a name="line.555"></a> -<span class="sourceLineNo">556</span> protected boolean includeRegionInSplit(final byte[] startKey, final byte [] endKey) {<a name="line.556"></a> -<span class="sourceLineNo">557</span> return true;<a name="line.557"></a> -<span class="sourceLineNo">558</span> }<a name="line.558"></a> -<span class="sourceLineNo">559</span><a name="line.559"></a> -<span class="sourceLineNo">560</span> /**<a name="line.560"></a> -<span class="sourceLineNo">561</span> * Allows subclasses to get the {@link Table}.<a name="line.561"></a> -<span class="sourceLineNo">562</span> *<a name="line.562"></a> -<span class="sourceLineNo">563</span> * @deprecated use {@link #getTable()}<a name="line.563"></a> -<span class="sourceLineNo">564</span> */<a name="line.564"></a> -<span class="sourceLineNo">565</span> @Deprecated<a name="line.565"></a> -<span class="sourceLineNo">566</span> protected HTable getHTable() {<a name="line.566"></a> -<span class="sourceLineNo">567</span> return (HTable) this.getTable();<a name="line.567"></a> -<span class="sourceLineNo">568</span> }<a name="line.568"></a> -<span class="sourceLineNo">569</span><a name="line.569"></a> -<span class="sourceLineNo">570</span> /**<a name="line.570"></a> -<span class="sourceLineNo">571</span> * Allows subclasses to get the {@link RegionLocator}.<a name="line.571"></a> -<span class="sourceLineNo">572</span> */<a name="line.572"></a> -<span class="sourceLineNo">573</span> protected RegionLocator getRegionLocator() {<a name="line.573"></a> -<span class="sourceLineNo">574</span> if (regionLocator == null) {<a name="line.574"></a> -<span class="sourceLineNo">575</span> throw new IllegalStateException(NOT_INITIALIZED);<a name="line.575"></a> -<span class="sourceLineNo">576</span> }<a name="line.576"></a> -<span class="sourceLineNo">577</span> return regionLocator;<a name="line.577"></a> -<span class="sourceLineNo">578</span> }<a name="line.578"></a> -<span class="sourceLineNo">579</span> <a name="line.579"></a> -<span class="sourceLineNo">580</span> /**<a name="line.580"></a> -<span class="sourceLineNo">581</span> * Allows subclasses to get the {@link Table}.<a name="line.581"></a> -<span class="sourceLineNo">582</span> */<a name="line.582"></a> -<span class="sourceLineNo">583</span> protected Table getTable() {<a name="line.583"></a> -<span class="sourceLineNo">584</span> if (table == null) {<a name="line.584"></a> -<span class="sourceLineNo">585</span> throw new IllegalStateException(NOT_INITIALIZED);<a name="line.585"></a> -<span class="sourceLineNo">586</span> }<a name="line.586"></a> -<span class="sourceLineNo">587</span> return table;<a name="line.587"></a> -<span class="sourceLineNo">588</span> }<a name="line.588"></a> -<span class="sourceLineNo">589</span><a name="line.589"></a> -<span class="sourceLineNo">590</span> /**<a name="line.590"></a> -<span class="sourceLineNo">591</span> * Allows subclasses to get the {@link Admin}.<a name="line.591"></a> +<span class="sourceLineNo">551</span> */<a name="line.551"></a> +<span class="sourceLineNo">552</span> protected boolean includeRegionInSplit(final byte[] startKey, final byte [] endKey) {<a name="line.552"></a> +<span class="sourceLineNo">553</span> return true;<a name="line.553"></a> +<span class="sourceLineNo">554</span> }<a name="line.554"></a> +<span class="sourceLineNo">555</span><a name="line.555"></a> +<span class="sourceLineNo">556</span> /**<a name="line.556"></a> +<span class="sourceLineNo">557</span> * Allows subclasses to get the {@link RegionLocator}.<a name="line.557"></a> +<span class="sourceLineNo">558</span> */<a name="line.558"></a> +<span class="sourceLineNo">559</span> protected RegionLocator getRegionLocator() {<a name="line.559"></a> +<span class="sourceLineNo">560</span> if (regionLocator == null) {<a name="line.560"></a> +<span class="sourceLineNo">561</span> throw new IllegalStateException(NOT_INITIALIZED);<a name="line.561"></a> +<span class="sourceLineNo">562</span> }<a name="line.562"></a> +<span class="sourceLineNo">563</span> return regionLocator;<a name="line.563"></a> +<span class="sourceLineNo">564</span> }<a name="line.564"></a> +<span class="sourceLineNo">565</span> <a name="line.565"></a> +<span class="sourceLineNo">566</span> /**<a name="line.566"></a> +<span class="sourceLineNo">567</span> * Allows subclasses to get the {@link Table}.<a name="line.567"></a> +<span class="sourceLineNo">568</span> */<a name="line.568"></a> +<span class="sourceLineNo">569</span> protected Table getTable() {<a name="line.569"></a> +<span class="sourceLineNo">570</span> if (table == null) {<a name="line.570"></a> +<span class="sourceLineNo">571</span> throw new IllegalStateException(NOT_INITIALIZED);<a name="line.571"></a> +<span class="sourceLineNo">572</span> }<a name="line.572"></a> +<span class="sourceLineNo">573</span> return table;<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> * Allows subclasses to get the {@link Admin}.<a name="line.577"></a> +<span class="sourceLineNo">578</span> */<a name="line.578"></a> +<span class="sourceLineNo">579</span> protected Admin getAdmin() {<a name="line.579"></a> +<span class="sourceLineNo">580</span> if (admin == null) {<a name="line.580"></a> +<span class="sourceLineNo">581</span> throw new IllegalStateException(NOT_INITIALIZED);<a name="line.581"></a> +<span class="sourceLineNo">582</span> }<a name="line.582"></a> +<span class="sourceLineNo">583</span> return admin;<a name="line.583"></a> +<span class="sourceLineNo">584</span> }<a name="line.584"></a> +<span class="sourceLineNo">585</span><a name="line.585"></a> +<span class="sourceLineNo">586</span> /**<a name="line.586"></a> +<span class="sourceLineNo">587</span> * Allows subclasses to initialize the table information.<a name="line.587"></a> +<span class="sourceLineNo">588</span> *<a name="line.588"></a> +<span class="sourceLineNo">589</span> * @param connection The Connection to the HBase cluster. MUST be unmanaged. We will close.<a name="line.589"></a> +<span class="sourceLineNo">590</span> * @param tableName The {@link TableName} of the table to process. <a name="line.590"></a> +<span class="sourceLineNo">591</span> * @throws IOException <a name="line.591"></a> <span class="sourceLineNo">592</span> */<a name="line.592"></a> -<span class="sourceLineNo">593</span> protected Admin getAdmin() {<a name="line.593"></a> -<span class="sourceLineNo">594</span> if (admin == null) {<a name="line.594"></a> -<span class="sourceLineNo">595</span> throw new IllegalStateException(NOT_INITIALIZED);<a name="line.595"></a> -<span class="sourceLineNo">596</span> }<a name="line.596"></a> -<span class="sourceLineNo">597</span> return admin;<a name="line.597"></a> -<span class="sourceLineNo">598</span> }<a name="line.598"></a> -<span class="sourceLineNo">599</span><a name="line.599"></a> -<span class="sourceLineNo">600</span> /**<a name="line.600"></a> -<span class="sourceLineNo">601</span> * Allows subclasses to set the {@link HTable}.<a name="line.601"></a> -<span class="sourceLineNo">602</span> *<a name="line.602"></a> -<span class="sourceLineNo">603</span> * Will attempt to reuse the underlying Connection for our own needs, including<a name="line.603"></a> -<span class="sourceLineNo">604</span> * retreiving an Admin interface to the HBase cluster.<a name="line.604"></a> -<span class="sourceLineNo">605</span> *<a name="line.605"></a> -<span class="sourceLineNo">606</span> * @param table The table to get the data from.<a name="line.606"></a> -<span class="sourceLineNo">607</span> * @throws IOException <a name="line.607"></a> -<span class="sourceLineNo">608</span> * @deprecated Use {@link #initializeTable(Connection, TableName)} instead.<a name="line.608"></a> -<span class="sourceLineNo">609</span> */<a name="line.609"></a> -<span class="sourceLineNo">610</span> @Deprecated<a name="line.610"></a> -<span class="sourceLineNo">611</span> protected void setHTable(HTable table) throws IOException {<a name="line.611"></a> -<span class="sourceLineNo">612</span> this.table = table;<a name="line.612"></a> -<span class="sourceLineNo">613</span> this.connection = table.getConnection();<a name="line.613"></a> -<span class="sourceLineNo">614</span> this.regionLocator = table.getRegionLocator();<a name="line.614"></a> -<span class="sourceLineNo">615</span> this.admin = this.connection.getAdmin();<a name="line.615"></a> -<span class="sourceLineNo">616</span> }<a name="line.616"></a> -<span class="sourceLineNo">617</span><a name="line.617"></a> -<span class="sourceLineNo">618</span> /**<a name="line.618"></a> -<span class="sourceLineNo">619</span> * Allows subclasses to initialize the table information.<a name="line.619"></a> -<span class="sourceLineNo">620</span> *<a name="line.620"></a> -<span class="sourceLineNo">621</span> * @param connection The Connection to the HBase cluster. MUST be unmanaged. We will close.<a name="line.621"></a> -<span class="sourceLineNo">622</span> * @param tableName The {@link TableName} of the table to process. <a name="line.622"></a> -<span class="sourceLineNo">623</span> * @throws IOException <a name="line.623"></a> -<span class="sourceLineNo">624</span> */<a name="line.624"></a> -<span class="sourceLineNo">625</span> protected void initializeTable(Connection connection, TableName tableName) throws IOException {<a name="line.625"></a> -<span class="sourceLineNo">626</span> if (this.table != null || this.connection != null) {<a name="line.626"></a> -<span class="sourceLineNo">627</span> LOG.warn("initializeTable called multiple times. Overwriting connection and table " +<a name="line.627"></a> -<span class="sourceLineNo">628</span> "reference; TableInputFormatBase will not close these old references when done.");<a name="line.628"></a> -<span class="sourceLineNo">629</span> }<a name="line.629"></a> -<span class="sourceLineNo">630</span> this.table = connection.getTable(tableName);<a name="line.630"></a> -<span class="sourceLineNo">631</span> this.regionLocator = connection.getRegionLocator(tableName);<a name="line.631"></a> -<span class="sourceLineNo">632</span> this.admin = connection.getAdmin();<a name="line.632"></a> -<span class="sourceLineNo">633</span> this.connection = connection;<a name="line.633"></a> -<span class="sourceLineNo">634</span> }<a name="line.634"></a> -<span class="sourceLineNo">635</span><a name="line.635"></a> -<span class="sourceLineNo">636</span> /**<a name="line.636"></a> -<span class="sourceLineNo">637</span> * Gets the scan defining the actual details like columns etc.<a name="line.637"></a> -<span class="sourceLineNo">638</span> *<a name="line.638"></a> -<span class="sourceLineNo">639</span> * @return The internal scan instance.<a name="line.639"></a> -<span class="sourceLineNo">640</span> */<a name="line.640"></a> -<span class="sourceLineNo">641</span> public Scan getScan() {<a name="line.641"></a> -<span class="sourceLineNo">642</span> if (this.scan == null) this.scan = new Scan();<a name="line.642"></a> -<span class="sourceLineNo">643</span> return scan;<a name="line.643"></a> -<span class="sourceLineNo">644</span> }<a name="line.644"></a> -<span class="sourceLineNo">645</span><a name="line.645"></a> -<span class="sourceLineNo">646</span> /**<a name="line.646"></a> -<span class="sourceLineNo">647</span> * Sets the scan defining the actual details like columns etc.<a name="line.647"></a> -<span class="sourceLineNo">648</span> *<a name="line.648"></a> -<span class="sourceLineNo">649</span> * @param scan The scan to set.<a name="line.649"></a> -<span class="sourceLineNo">650</span> */<a name="line.650"></a> -<span class="sourceLineNo">651</span> public void setScan(Scan scan) {<a name="line.651"></a> -<span class="sourceLineNo">652</span> this.scan = scan;<a name="line.652"></a> -<span class="sourceLineNo">653</span> }<a name="line.653"></a> -<span class="sourceLineNo">654</span><a name="line.654"></a> -<span class="sourceLineNo">655</span> /**<a name="line.655"></a> -<span class="sourceLineNo">656</span> * Allows subclasses to set the {@link TableRecordReader}.<a name="line.656"></a> -<span class="sourceLineNo">657</span> *<a name="line.657"></a> -<span class="sourceLineNo">658</span> * @param tableRecordReader A different {@link TableRecordReader}<a name="line.658"></a> -<span class="sourceLineNo">659</span> * implementation.<a name="line.659"></a> -<span class="sourceLineNo">660</span> */<a name="line.660"></a> -<span class="sourceLineNo">661</span> protected void setTableRecordReader(TableRecordReader tableRecordReader) {<a name="line.661"></a> -<span class="sourceLineNo">662</span> this.tableRecordReader = tableRecordReader;<a name="line.662"></a> +<span class="sourceLineNo">593</span> protected void initializeTable(Connection connection, TableName tableName) throws IOException {<a name="line.593"></a> +<span class="sourceLineNo">594</span> if (this.table != null || this.connection != null) {<a name="line.594"></a> +<span class="sourceLineNo">595</span> LOG.warn("initializeTable called multiple times. Overwriting connection and table " +<a name="line.595"></a> +<span class="sourceLineNo">596</span> "reference; TableInputFormatBase will not close these old references when done.");<a name="line.596"></a> +<span class="sourceLineNo">597</span> }<a name="line.597"></a> +<span class="sourceLineNo">598</span> this.table = connection.getTable(tableName);<a name="line.598"></a> +<span class="sourceLineNo">599</span> this.regionLocator = connection.getRegionLocator(tableName);<a name="line.599"></a> +<span class="sourceLineNo">600</span> this.admin = connection.getAdmin();<a name="line.600"></a> +<span class="sourceLineNo">601</span> this.connection = connection;<a name="line.601"></a> +<span class="sourceLineNo">602</span> }<a name="line.602"></a> +<span class="sourceLineNo">603</span><a name="line.603"></a> +<span class="sourceLineNo">604</span> /**<a name="line.604"></a> +<span class="sourceLineNo">605</span> * Gets the scan defining the actual details like columns etc.<a name="line.605"></a> +<span class="sourceLineNo">606</span> *<a name="line.606"></a> +<span class="sourceLineNo">607</span> * @return The internal scan instance.<a name="line.607"></a> +<span class="sourceLineNo">608</span> */<a name="line.608"></a> +<span class="sourceLineNo">609</span> public Scan getScan() {<a name="line.609"></a> +<span class="sourceLineNo">610</span> if (this.scan == null) this.scan = new Scan();<a name="line.610"></a> +<span class="sourceLineNo">611</span> return scan;<a name="line.611"></a> +<span class="sourceLineNo">612</span> }<a name="line.612"></a> +<span class="sourceLineNo">613</span><a name="line.613"></a> +<span class="sourceLineNo">614</span> /**<a name="line.614"></a> +<span class="sourceLineNo">615</span> * Sets the scan defining the actual details like columns etc.<a name="line.615"></a> +<span class="sourceLineNo">616</span> *<a name="line.616"></a> +<span class="sourceLineNo">617</span> * @param scan The scan to set.<a name="line.617"></a> +<span class="sourceLineNo">618</span> */<a name="line.618"></a> +<span class="sourceLineNo">619</span> public void setScan(Scan scan) {<a name="line.619"></a> +<span class="sourceLineNo">620</span> this.scan = scan;<a name="line.620"></a> +<span class="sourceLineNo">621</span> }<a name="line.621"></a> +<span class="sourceLineNo">622</span><a name="line.622"></a> +<span class="sourceLineNo">623</span> /**<a name="line.623"></a> +<span class="sourceLineNo">624</span> * Allows subclasses to set the {@link TableRecordReader}.<a name="line.624"></a> +<span class="sourceLineNo">625</span> *<a name="line.625"></a> +<span class="sourceLineNo">626</span> * @param tableRecordReader A different {@link TableRecordReader}<a name="line.626"></a> +<span class="sourceLineNo">627</span> * implementation.<a name="line.627"></a> +<span class="sourceLineNo">628</span> */<a name="line.628"></a> +<span class="sourceLineNo">629</span> protected void setTableRecordReader(TableRecordReader tableRecordReader) {<a name="line.629"></a> +<span class="sourceLineNo">630</span> this.tableRecordReader = tableRecordReader;<a name="line.630"></a> +<span class="sourceLineNo">631</span> }<a name="line.631"></a> +<span class="sourceLineNo">632</span> <a name="line.632"></a> +<span class="sourceLineNo">633</span> /**<a name="line.633"></a> +<span class="sourceLineNo">634</span> * Handle subclass specific set up.<a name="line.634"></a> +<span class="sourceLineNo">635</span> * Each of the entry points used by the MapReduce framework,<a name="line.635"></a> +<span class="sourceLineNo">636</span> * {@link #createRecordReader(InputSplit, TaskAttemptContext)} and {@link #getSplits(JobContext)},<a name="line.636"></a> +<span class="sourceLineNo">637</span> * will call {@link #initialize(JobContext)} as a convenient centralized location to handle<a name="line.637"></a> +<span class="sourceLineNo">638</span> * retrieving the necessary configuration information and calling<a name="line.638"></a> +<span class="sourceLineNo">639</span> * {@link #initializeTable(Connection, TableName)}.<a name="line.639"></a> +<span class="sourceLineNo">640</span> *<a name="line.640"></a> +<span class="sourceLineNo">641</span> * Subclasses should implement their initialize call such that it is safe to call multiple times.<a name="line.641"></a> +<span class="sourceLineNo">642</span> * The current TableInputFormatBase implementation relies on a non-null table reference to decide<a name="line.642"></a> +<span class="sourceLineNo">643</span> * if an initialize call is needed, but this behavior may change in the future. In particular,<a name="line.643"></a> +<span class="sourceLineNo">644</span> * it is critical that initializeTable not be called multiple times since this will leak<a name="line.644"></a> +<span class="sourceLineNo">645</span> * Connection instances.<a name="line.645"></a> +<span class="sourceLineNo">646</span> *<a name="line.646"></a> +<span class="sourceLineNo">647</span> */<a name="line.647"></a> +<span class="sourceLineNo">648</span> protected void initialize(JobContext context) throws IOException {<a name="line.648"></a> +<span class="sourceLineNo">649</span> }<a name="line.649"></a> +<span class="sourceLineNo">650</span><a name="line.650"></a> +<span class="sourceLineNo">651</span> /**<a name="line.651"></a> +<span class="sourceLineNo">652</span> * Close the Table and related objects that were initialized via<a name="line.652"></a> +<span class="sourceLineNo">653</span> * {@link #initializeTable(Connection, TableName)}.<a name="line.653"></a> +<span class="sourceLineNo">654</span> *<a name="line.654"></a> +<span class="sourceLineNo">655</span> * @throws IOException<a name="line.655"></a> +<span class="sourceLineNo">656</span> */<a name="line.656"></a> +<span class="sourceLineNo">657</span> protected void closeTable() throws IOException {<a name="line.657"></a> +<span class="sourceLineNo">658</span> close(admin, table, regionLocator, connection);<a name="line.658"></a> +<span class="sourceLineNo">659</span> admin = null;<a name="line.659"></a> +<span class="sourceLineNo">660</span> table = null;<a name="line.660"></a> +<span class="sourceLineNo">661</span> regionLocator = null;<a name="line.661"></a> +<span class="sourceLineNo">662</span> connection = null;<a name="line.662"></a> <span class="sourceLineNo">663</span> }<a name="line.663"></a> -<span class="sourceLineNo">664</span> <a name="line.664"></a> -<span class="sourceLineNo">665</span> /**<a name="line.665"></a> -<span class="sourceLineNo">666</span> * Handle subclass specific set up.<a name="line.666"></a> -<span class="sourceLineNo">667</span> * Each of the entry points used by the MapReduce framework,<a name="line.667"></a> -<span class="sourceLineNo">668</span> * {@link #createRecordReader(InputSplit, TaskAttemptContext)} and {@link #getSplits(JobContext)},<a name="line.668"></a> -<span class="sourceLineNo">669</span> * will call {@link #initialize(JobContext)} as a convenient centralized location to handle<a name="line.669"></a> -<span class="sourceLineNo">670</span> * retrieving the necessary configuration information and calling<a name="line.670"></a> -<span class="sourceLineNo">671</span> * {@link #initializeTable(Connection, TableName)}.<a name="line.671"></a> -<span class="sourceLineNo">672</span> *<a name="line.672"></a> -<span class="sourceLineNo">673</span> * Subclasses should implement their initialize call such that it is safe to call multiple times.<a name="line.673"></a> -<span class="sourceLineNo">674</span> * The current TableInputFormatBase implementation relies on a non-null table reference to decide<a name="line.674"></a> -<span class="sourceLineNo">675</span> * if an initialize call is needed, but this behavior may change in the future. In particular,<a name="line.675"></a> -<span class="sourceLineNo">676</span> * it is critical that initializeTable not be called multiple times since this will leak<a name="line.676"></a> -<span class="sourceLineNo">677</span> * Connection instances.<a name="line.677"></a> -<span class="sourceLineNo">678</span> *<a name="line.678"></a> -<span class="sourceLineNo">679</span> */<a name="line.679"></a> -<span class="sourceLineNo">680</span> protected void initialize(JobContext context) throws IOException {<a name="line.680"></a> -<span class="sourceLineNo">681</span> }<a name="line.681"></a> -<span class="sourceLineNo">682</span><a name="line.682"></a> -<span class="sourceLineNo">683</span> /**<a name="line.683"></a> -<span class="sourceLineNo">684</span> * Close the Table and related objects that were initialized via<a name="line.684"></a> -<span class="sourceLineNo">685</span> * {@link #initializeTable(Connection, TableName)}.<a name="line.685"></a> -<span class="sourceLineNo">686</span> *<a name="line.686"></a> -<span class="sourceLineNo">687</span> * @throws IOException<a name="line.687"></a> -<span class="sourceLineNo">688</span> */<a name="line.688"></a> -<span class="sourceLineNo">689</span> protected void closeTable() throws IOException {<a name="line.689"></a> -<span class="sourceLineNo">690</span> close(admin, table, regionLocator, connection);<a name="line.690"></a> -<span class="sourceLineNo">691</span> admin = null;<a name="line.691"></a> -<span class="sourceLineNo">692</span> table = null;<a name="line.692"></a> -<span class="sourceLineNo">693</span> regionLocator = null;<a name="line.693"></a> -<span class="sourceLineNo">694</span> connection = null;<a name="line.694"></a> -<span class="sourceLineNo">695</span> }<a name="line.695"></a> -<span class="sourceLineNo">696</span><a name="line.696"></a> -<span class="sourceLineNo">697</span> private void close(Closeable... closables) throws IOException {<a name="line.697"></a> -<span class="sourceLineNo">698</span> for (Closeable c : closables) {<a name="line.698"></a> -<span class="sourceLineNo">699</span> if(c != null) { c.close(); }<a name="line.699"></a> -<span class="sourceLineNo">700</span> }<a name="line.700"></a> -<span class="sourceLineNo">701</span> }<a name="line.701"></a> -<span class="sourceLineNo">702</span><a name="line.702"></a> -<span class="sourceLineNo">703</span>}<a name="line.703"></a> +<span class="sourceLineNo">664</span><a name="line.664"></a> +<span class="sourceLineNo">665</span> private void close(Closeable... closables) throws IOException {<a name="line.665"></a> +<span class="sourceLineNo">666</span> for (Closeable c : closables) {<a name="line.666"></a> +<span class="sourceLineNo">667</span> if(c != null) { c.close(); }<a name="line.667"></a> +<span class="sourceLineNo">668</span> }<a name="line.668"></a> +<span class="sourceLineNo">669</span> }<a name="line.669"></a> +<span class="sourceLineNo">670</span><a name="line.670"></a> +<span class="sourceLineNo">671</span>}<a name="line.671"></a>
