http://git-wip-us.apache.org/repos/asf/hbase-site/blob/387c1112/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html index 5478df1..11bec12 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html @@ -86,19 +86,19 @@ <span class="sourceLineNo">078</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.78"></a> <span class="sourceLineNo">079</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.79"></a> <span class="sourceLineNo">080</span>import org.apache.hadoop.hbase.util.FSUtils;<a name="line.80"></a> -<span class="sourceLineNo">081</span>import org.apache.hadoop.io.NullWritable;<a name="line.81"></a> -<span class="sourceLineNo">082</span>import org.apache.hadoop.io.SequenceFile;<a name="line.82"></a> -<span class="sourceLineNo">083</span>import org.apache.hadoop.io.Text;<a name="line.83"></a> -<span class="sourceLineNo">084</span>import org.apache.hadoop.mapreduce.Job;<a name="line.84"></a> -<span class="sourceLineNo">085</span>import org.apache.hadoop.mapreduce.OutputCommitter;<a name="line.85"></a> -<span class="sourceLineNo">086</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.86"></a> -<span class="sourceLineNo">087</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.87"></a> -<span class="sourceLineNo">088</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.88"></a> -<span class="sourceLineNo">089</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter;<a name="line.89"></a> -<span class="sourceLineNo">090</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.90"></a> -<span class="sourceLineNo">091</span>import org.apache.hadoop.mapreduce.lib.partition.TotalOrderPartitioner;<a name="line.91"></a> -<span class="sourceLineNo">092</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.92"></a> -<span class="sourceLineNo">093</span><a name="line.93"></a> +<span class="sourceLineNo">081</span>import org.apache.hadoop.hbase.util.MapReduceCell;<a name="line.81"></a> +<span class="sourceLineNo">082</span>import org.apache.hadoop.io.NullWritable;<a name="line.82"></a> +<span class="sourceLineNo">083</span>import org.apache.hadoop.io.SequenceFile;<a name="line.83"></a> +<span class="sourceLineNo">084</span>import org.apache.hadoop.io.Text;<a name="line.84"></a> +<span class="sourceLineNo">085</span>import org.apache.hadoop.mapreduce.Job;<a name="line.85"></a> +<span class="sourceLineNo">086</span>import org.apache.hadoop.mapreduce.OutputCommitter;<a name="line.86"></a> +<span class="sourceLineNo">087</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.87"></a> +<span class="sourceLineNo">088</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.88"></a> +<span class="sourceLineNo">089</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.89"></a> +<span class="sourceLineNo">090</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter;<a name="line.90"></a> +<span class="sourceLineNo">091</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.91"></a> +<span class="sourceLineNo">092</span>import org.apache.hadoop.mapreduce.lib.partition.TotalOrderPartitioner;<a name="line.92"></a> +<span class="sourceLineNo">093</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.93"></a> <span class="sourceLineNo">094</span>import org.apache.hadoop.hbase.shaded.com.google.common.annotations.VisibleForTesting;<a name="line.94"></a> <span class="sourceLineNo">095</span><a name="line.95"></a> <span class="sourceLineNo">096</span>/**<a name="line.96"></a> @@ -241,111 +241,111 @@ <span class="sourceLineNo">233</span> private final Map<byte[], WriterLength> writers =<a name="line.233"></a> <span class="sourceLineNo">234</span> new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.234"></a> <span class="sourceLineNo">235</span> private byte[] previousRow = HConstants.EMPTY_BYTE_ARRAY;<a name="line.235"></a> -<span class="sourceLineNo">236</span> private final byte[] now = Bytes.toBytes(EnvironmentEdgeManager.currentTime());<a name="line.236"></a> +<span class="sourceLineNo">236</span> private final long now = EnvironmentEdgeManager.currentTime();<a name="line.236"></a> <span class="sourceLineNo">237</span> private boolean rollRequested = false;<a name="line.237"></a> <span class="sourceLineNo">238</span><a name="line.238"></a> <span class="sourceLineNo">239</span> @Override<a name="line.239"></a> <span class="sourceLineNo">240</span> public void write(ImmutableBytesWritable row, V cell)<a name="line.240"></a> <span class="sourceLineNo">241</span> throws IOException {<a name="line.241"></a> -<span class="sourceLineNo">242</span> KeyValue kv = KeyValueUtil.ensureKeyValue(cell);<a name="line.242"></a> -<span class="sourceLineNo">243</span><a name="line.243"></a> -<span class="sourceLineNo">244</span> // null input == user explicitly wants to flush<a name="line.244"></a> -<span class="sourceLineNo">245</span> if (row == null && kv == null) {<a name="line.245"></a> -<span class="sourceLineNo">246</span> rollWriters();<a name="line.246"></a> -<span class="sourceLineNo">247</span> return;<a name="line.247"></a> -<span class="sourceLineNo">248</span> }<a name="line.248"></a> -<span class="sourceLineNo">249</span><a name="line.249"></a> -<span class="sourceLineNo">250</span> byte[] rowKey = CellUtil.cloneRow(kv);<a name="line.250"></a> -<span class="sourceLineNo">251</span> long length = kv.getLength();<a name="line.251"></a> -<span class="sourceLineNo">252</span> byte[] family = CellUtil.cloneFamily(kv);<a name="line.252"></a> -<span class="sourceLineNo">253</span> byte[] tableNameBytes = null;<a name="line.253"></a> -<span class="sourceLineNo">254</span> if (writeMultipleTables) {<a name="line.254"></a> -<span class="sourceLineNo">255</span> tableNameBytes = MultiTableHFileOutputFormat.getTableName(row.get());<a name="line.255"></a> -<span class="sourceLineNo">256</span> if (!allTableNames.contains(Bytes.toString(tableNameBytes))) {<a name="line.256"></a> -<span class="sourceLineNo">257</span> throw new IllegalArgumentException("TableName '" + Bytes.toString(tableNameBytes) +<a name="line.257"></a> -<span class="sourceLineNo">258</span> "' not" + " expected");<a name="line.258"></a> -<span class="sourceLineNo">259</span> }<a name="line.259"></a> -<span class="sourceLineNo">260</span> } else {<a name="line.260"></a> -<span class="sourceLineNo">261</span> tableNameBytes = writeTableNames.getBytes(StandardCharsets.UTF_8);<a name="line.261"></a> -<span class="sourceLineNo">262</span> }<a name="line.262"></a> -<span class="sourceLineNo">263</span> byte[] tableAndFamily = getTableNameSuffixedWithFamily(tableNameBytes, family);<a name="line.263"></a> -<span class="sourceLineNo">264</span> WriterLength wl = this.writers.get(tableAndFamily);<a name="line.264"></a> -<span class="sourceLineNo">265</span><a name="line.265"></a> -<span class="sourceLineNo">266</span> // If this is a new column family, verify that the directory exists<a name="line.266"></a> -<span class="sourceLineNo">267</span> if (wl == null) {<a name="line.267"></a> -<span class="sourceLineNo">268</span> Path writerPath = null;<a name="line.268"></a> -<span class="sourceLineNo">269</span> if (writeMultipleTables) {<a name="line.269"></a> -<span class="sourceLineNo">270</span> writerPath = new Path(outputDir, new Path(Bytes.toString(tableNameBytes), Bytes<a name="line.270"></a> -<span class="sourceLineNo">271</span> .toString(family)));<a name="line.271"></a> -<span class="sourceLineNo">272</span> }<a name="line.272"></a> -<span class="sourceLineNo">273</span> else {<a name="line.273"></a> -<span class="sourceLineNo">274</span> writerPath = new Path(outputDir, Bytes.toString(family));<a name="line.274"></a> -<span class="sourceLineNo">275</span> }<a name="line.275"></a> -<span class="sourceLineNo">276</span> fs.mkdirs(writerPath);<a name="line.276"></a> -<span class="sourceLineNo">277</span> configureStoragePolicy(conf, fs, tableAndFamily, writerPath);<a name="line.277"></a> -<span class="sourceLineNo">278</span> }<a name="line.278"></a> -<span class="sourceLineNo">279</span><a name="line.279"></a> -<span class="sourceLineNo">280</span> // If any of the HFiles for the column families has reached<a name="line.280"></a> -<span class="sourceLineNo">281</span> // maxsize, we need to roll all the writers<a name="line.281"></a> -<span class="sourceLineNo">282</span> if (wl != null && wl.written + length >= maxsize) {<a name="line.282"></a> -<span class="sourceLineNo">283</span> this.rollRequested = true;<a name="line.283"></a> -<span class="sourceLineNo">284</span> }<a name="line.284"></a> -<span class="sourceLineNo">285</span><a name="line.285"></a> -<span class="sourceLineNo">286</span> // This can only happen once a row is finished though<a name="line.286"></a> -<span class="sourceLineNo">287</span> if (rollRequested && Bytes.compareTo(this.previousRow, rowKey) != 0) {<a name="line.287"></a> -<span class="sourceLineNo">288</span> rollWriters();<a name="line.288"></a> -<span class="sourceLineNo">289</span> }<a name="line.289"></a> -<span class="sourceLineNo">290</span><a name="line.290"></a> -<span class="sourceLineNo">291</span> // create a new WAL writer, if necessary<a name="line.291"></a> -<span class="sourceLineNo">292</span> if (wl == null || wl.writer == null) {<a name="line.292"></a> -<span class="sourceLineNo">293</span> if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.293"></a> -<span class="sourceLineNo">294</span> HRegionLocation loc = null;<a name="line.294"></a> -<span class="sourceLineNo">295</span><a name="line.295"></a> -<span class="sourceLineNo">296</span> String tableName = Bytes.toString(tableNameBytes);<a name="line.296"></a> -<span class="sourceLineNo">297</span> if (tableName != null) {<a name="line.297"></a> -<span class="sourceLineNo">298</span> try (Connection connection = ConnectionFactory.createConnection(conf);<a name="line.298"></a> -<span class="sourceLineNo">299</span> RegionLocator locator =<a name="line.299"></a> -<span class="sourceLineNo">300</span> connection.getRegionLocator(TableName.valueOf(tableName))) {<a name="line.300"></a> -<span class="sourceLineNo">301</span> loc = locator.getRegionLocation(rowKey);<a name="line.301"></a> -<span class="sourceLineNo">302</span> } catch (Throwable e) {<a name="line.302"></a> -<span class="sourceLineNo">303</span> LOG.warn("There's something wrong when locating rowkey: " +<a name="line.303"></a> -<span class="sourceLineNo">304</span> Bytes.toString(rowKey) + " for tablename: " + tableName, e);<a name="line.304"></a> -<span class="sourceLineNo">305</span> loc = null;<a name="line.305"></a> -<span class="sourceLineNo">306</span> } }<a name="line.306"></a> -<span class="sourceLineNo">307</span><a name="line.307"></a> -<span class="sourceLineNo">308</span> if (null == loc) {<a name="line.308"></a> -<span class="sourceLineNo">309</span> if (LOG.isTraceEnabled()) {<a name="line.309"></a> -<span class="sourceLineNo">310</span> LOG.trace("failed to get region location, so use default writer for rowkey: " +<a name="line.310"></a> -<span class="sourceLineNo">311</span> Bytes.toString(rowKey));<a name="line.311"></a> -<span class="sourceLineNo">312</span> }<a name="line.312"></a> -<span class="sourceLineNo">313</span> wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.313"></a> -<span class="sourceLineNo">314</span> } else {<a name="line.314"></a> -<span class="sourceLineNo">315</span> if (LOG.isDebugEnabled()) {<a name="line.315"></a> -<span class="sourceLineNo">316</span> LOG.debug("first rowkey: [" + Bytes.toString(rowKey) + "]");<a name="line.316"></a> -<span class="sourceLineNo">317</span> }<a name="line.317"></a> -<span class="sourceLineNo">318</span> InetSocketAddress initialIsa =<a name="line.318"></a> -<span class="sourceLineNo">319</span> new InetSocketAddress(loc.getHostname(), loc.getPort());<a name="line.319"></a> -<span class="sourceLineNo">320</span> if (initialIsa.isUnresolved()) {<a name="line.320"></a> -<span class="sourceLineNo">321</span> if (LOG.isTraceEnabled()) {<a name="line.321"></a> -<span class="sourceLineNo">322</span> LOG.trace("failed to resolve bind address: " + loc.getHostname() + ":"<a name="line.322"></a> -<span class="sourceLineNo">323</span> + loc.getPort() + ", so use default writer");<a name="line.323"></a> -<span class="sourceLineNo">324</span> }<a name="line.324"></a> -<span class="sourceLineNo">325</span> wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.325"></a> -<span class="sourceLineNo">326</span> } else {<a name="line.326"></a> -<span class="sourceLineNo">327</span> if (LOG.isDebugEnabled()) {<a name="line.327"></a> -<span class="sourceLineNo">328</span> LOG.debug("use favored nodes writer: " + initialIsa.getHostString());<a name="line.328"></a> -<span class="sourceLineNo">329</span> }<a name="line.329"></a> -<span class="sourceLineNo">330</span> wl = getNewWriter(tableNameBytes, family, conf, new InetSocketAddress[] { initialIsa<a name="line.330"></a> -<span class="sourceLineNo">331</span> });<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> } else {<a name="line.334"></a> -<span class="sourceLineNo">335</span> wl = getNewWriter(tableNameBytes, family, conf, null);<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> // we now have the proper WAL writer. full steam ahead<a name="line.339"></a> -<span class="sourceLineNo">340</span> kv.updateLatestStamp(this.now);<a name="line.340"></a> +<span class="sourceLineNo">242</span> Cell kv = cell;<a name="line.242"></a> +<span class="sourceLineNo">243</span> // null input == user explicitly wants to flush<a name="line.243"></a> +<span class="sourceLineNo">244</span> if (row == null && kv == null) {<a name="line.244"></a> +<span class="sourceLineNo">245</span> rollWriters();<a name="line.245"></a> +<span class="sourceLineNo">246</span> return;<a name="line.246"></a> +<span class="sourceLineNo">247</span> }<a name="line.247"></a> +<span class="sourceLineNo">248</span><a name="line.248"></a> +<span class="sourceLineNo">249</span> byte[] rowKey = CellUtil.cloneRow(kv);<a name="line.249"></a> +<span class="sourceLineNo">250</span> int length = (CellUtil.estimatedSerializedSizeOf(kv)) - Bytes.SIZEOF_INT;<a name="line.250"></a> +<span class="sourceLineNo">251</span> byte[] family = CellUtil.cloneFamily(kv);<a name="line.251"></a> +<span class="sourceLineNo">252</span> byte[] tableNameBytes = null;<a name="line.252"></a> +<span class="sourceLineNo">253</span> if (writeMultipleTables) {<a name="line.253"></a> +<span class="sourceLineNo">254</span> tableNameBytes = MultiTableHFileOutputFormat.getTableName(row.get());<a name="line.254"></a> +<span class="sourceLineNo">255</span> if (!allTableNames.contains(Bytes.toString(tableNameBytes))) {<a name="line.255"></a> +<span class="sourceLineNo">256</span> throw new IllegalArgumentException("TableName '" + Bytes.toString(tableNameBytes) +<a name="line.256"></a> +<span class="sourceLineNo">257</span> "' not" + " expected");<a name="line.257"></a> +<span class="sourceLineNo">258</span> }<a name="line.258"></a> +<span class="sourceLineNo">259</span> } else {<a name="line.259"></a> +<span class="sourceLineNo">260</span> tableNameBytes = writeTableNames.getBytes(StandardCharsets.UTF_8);<a name="line.260"></a> +<span class="sourceLineNo">261</span> }<a name="line.261"></a> +<span class="sourceLineNo">262</span> byte[] tableAndFamily = getTableNameSuffixedWithFamily(tableNameBytes, family);<a name="line.262"></a> +<span class="sourceLineNo">263</span> WriterLength wl = this.writers.get(tableAndFamily);<a name="line.263"></a> +<span class="sourceLineNo">264</span><a name="line.264"></a> +<span class="sourceLineNo">265</span> // If this is a new column family, verify that the directory exists<a name="line.265"></a> +<span class="sourceLineNo">266</span> if (wl == null) {<a name="line.266"></a> +<span class="sourceLineNo">267</span> Path writerPath = null;<a name="line.267"></a> +<span class="sourceLineNo">268</span> if (writeMultipleTables) {<a name="line.268"></a> +<span class="sourceLineNo">269</span> writerPath = new Path(outputDir, new Path(Bytes.toString(tableNameBytes), Bytes<a name="line.269"></a> +<span class="sourceLineNo">270</span> .toString(family)));<a name="line.270"></a> +<span class="sourceLineNo">271</span> }<a name="line.271"></a> +<span class="sourceLineNo">272</span> else {<a name="line.272"></a> +<span class="sourceLineNo">273</span> writerPath = new Path(outputDir, Bytes.toString(family));<a name="line.273"></a> +<span class="sourceLineNo">274</span> }<a name="line.274"></a> +<span class="sourceLineNo">275</span> fs.mkdirs(writerPath);<a name="line.275"></a> +<span class="sourceLineNo">276</span> configureStoragePolicy(conf, fs, tableAndFamily, writerPath);<a name="line.276"></a> +<span class="sourceLineNo">277</span> }<a name="line.277"></a> +<span class="sourceLineNo">278</span><a name="line.278"></a> +<span class="sourceLineNo">279</span> // If any of the HFiles for the column families has reached<a name="line.279"></a> +<span class="sourceLineNo">280</span> // maxsize, we need to roll all the writers<a name="line.280"></a> +<span class="sourceLineNo">281</span> if (wl != null && wl.written + length >= maxsize) {<a name="line.281"></a> +<span class="sourceLineNo">282</span> this.rollRequested = true;<a name="line.282"></a> +<span class="sourceLineNo">283</span> }<a name="line.283"></a> +<span class="sourceLineNo">284</span><a name="line.284"></a> +<span class="sourceLineNo">285</span> // This can only happen once a row is finished though<a name="line.285"></a> +<span class="sourceLineNo">286</span> if (rollRequested && Bytes.compareTo(this.previousRow, rowKey) != 0) {<a name="line.286"></a> +<span class="sourceLineNo">287</span> rollWriters();<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> // create a new WAL writer, if necessary<a name="line.290"></a> +<span class="sourceLineNo">291</span> if (wl == null || wl.writer == null) {<a name="line.291"></a> +<span class="sourceLineNo">292</span> if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.292"></a> +<span class="sourceLineNo">293</span> HRegionLocation loc = null;<a name="line.293"></a> +<span class="sourceLineNo">294</span><a name="line.294"></a> +<span class="sourceLineNo">295</span> String tableName = Bytes.toString(tableNameBytes);<a name="line.295"></a> +<span class="sourceLineNo">296</span> if (tableName != null) {<a name="line.296"></a> +<span class="sourceLineNo">297</span> try (Connection connection = ConnectionFactory.createConnection(conf);<a name="line.297"></a> +<span class="sourceLineNo">298</span> RegionLocator locator =<a name="line.298"></a> +<span class="sourceLineNo">299</span> connection.getRegionLocator(TableName.valueOf(tableName))) {<a name="line.299"></a> +<span class="sourceLineNo">300</span> loc = locator.getRegionLocation(rowKey);<a name="line.300"></a> +<span class="sourceLineNo">301</span> } catch (Throwable e) {<a name="line.301"></a> +<span class="sourceLineNo">302</span> LOG.warn("There's something wrong when locating rowkey: " +<a name="line.302"></a> +<span class="sourceLineNo">303</span> Bytes.toString(rowKey) + " for tablename: " + tableName, e);<a name="line.303"></a> +<span class="sourceLineNo">304</span> loc = null;<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> if (null == loc) {<a name="line.307"></a> +<span class="sourceLineNo">308</span> if (LOG.isTraceEnabled()) {<a name="line.308"></a> +<span class="sourceLineNo">309</span> LOG.trace("failed to get region location, so use default writer for rowkey: " +<a name="line.309"></a> +<span class="sourceLineNo">310</span> Bytes.toString(rowKey));<a name="line.310"></a> +<span class="sourceLineNo">311</span> }<a name="line.311"></a> +<span class="sourceLineNo">312</span> wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.312"></a> +<span class="sourceLineNo">313</span> } else {<a name="line.313"></a> +<span class="sourceLineNo">314</span> if (LOG.isDebugEnabled()) {<a name="line.314"></a> +<span class="sourceLineNo">315</span> LOG.debug("first rowkey: [" + Bytes.toString(rowKey) + "]");<a name="line.315"></a> +<span class="sourceLineNo">316</span> }<a name="line.316"></a> +<span class="sourceLineNo">317</span> InetSocketAddress initialIsa =<a name="line.317"></a> +<span class="sourceLineNo">318</span> new InetSocketAddress(loc.getHostname(), loc.getPort());<a name="line.318"></a> +<span class="sourceLineNo">319</span> if (initialIsa.isUnresolved()) {<a name="line.319"></a> +<span class="sourceLineNo">320</span> if (LOG.isTraceEnabled()) {<a name="line.320"></a> +<span class="sourceLineNo">321</span> LOG.trace("failed to resolve bind address: " + loc.getHostname() + ":"<a name="line.321"></a> +<span class="sourceLineNo">322</span> + loc.getPort() + ", so use default writer");<a name="line.322"></a> +<span class="sourceLineNo">323</span> }<a name="line.323"></a> +<span class="sourceLineNo">324</span> wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.324"></a> +<span class="sourceLineNo">325</span> } else {<a name="line.325"></a> +<span class="sourceLineNo">326</span> if (LOG.isDebugEnabled()) {<a name="line.326"></a> +<span class="sourceLineNo">327</span> LOG.debug("use favored nodes writer: " + initialIsa.getHostString());<a name="line.327"></a> +<span class="sourceLineNo">328</span> }<a name="line.328"></a> +<span class="sourceLineNo">329</span> wl = getNewWriter(tableNameBytes, family, conf, new InetSocketAddress[] { initialIsa<a name="line.329"></a> +<span class="sourceLineNo">330</span> });<a name="line.330"></a> +<span class="sourceLineNo">331</span> }<a name="line.331"></a> +<span class="sourceLineNo">332</span> }<a name="line.332"></a> +<span class="sourceLineNo">333</span> } else {<a name="line.333"></a> +<span class="sourceLineNo">334</span> wl = getNewWriter(tableNameBytes, family, conf, null);<a name="line.334"></a> +<span class="sourceLineNo">335</span> }<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> // we now have the proper WAL writer. full steam ahead<a name="line.338"></a> +<span class="sourceLineNo">339</span> // TODO : Currently in SettableTimeStamp but this will also move to ExtendedCell<a name="line.339"></a> +<span class="sourceLineNo">340</span> CellUtil.updateLatestStamp(cell, this.now);<a name="line.340"></a> <span class="sourceLineNo">341</span> wl.writer.append(kv);<a name="line.341"></a> <span class="sourceLineNo">342</span> wl.written += length;<a name="line.342"></a> <span class="sourceLineNo">343</span><a name="line.343"></a> @@ -586,332 +586,334 @@ <span class="sourceLineNo">578</span> configureIncrementalLoad(job, singleTableInfo, HFileOutputFormat2.class);<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> static void configureIncrementalLoad(Job job, List<TableInfo> multiTableInfo, Class<? extends OutputFormat<?, ?>> cls) throws IOException {<a name="line.581"></a> -<span class="sourceLineNo">582</span> Configuration conf = job.getConfiguration();<a name="line.582"></a> -<span class="sourceLineNo">583</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.583"></a> -<span class="sourceLineNo">584</span> job.setOutputValueClass(KeyValue.class);<a name="line.584"></a> -<span class="sourceLineNo">585</span> job.setOutputFormatClass(cls);<a name="line.585"></a> -<span class="sourceLineNo">586</span><a name="line.586"></a> -<span class="sourceLineNo">587</span> if (multiTableInfo.stream().distinct().count() != multiTableInfo.size()) {<a name="line.587"></a> -<span class="sourceLineNo">588</span> throw new IllegalArgumentException("Duplicate entries found in TableInfo argument");<a name="line.588"></a> -<span class="sourceLineNo">589</span> }<a name="line.589"></a> -<span class="sourceLineNo">590</span> boolean writeMultipleTables = false;<a name="line.590"></a> -<span class="sourceLineNo">591</span> if (MultiTableHFileOutputFormat.class.equals(cls)) {<a name="line.591"></a> -<span class="sourceLineNo">592</span> writeMultipleTables = true;<a name="line.592"></a> -<span class="sourceLineNo">593</span> conf.setBoolean(MULTI_TABLE_HFILEOUTPUTFORMAT_CONF_KEY, true);<a name="line.593"></a> -<span class="sourceLineNo">594</span> }<a name="line.594"></a> -<span class="sourceLineNo">595</span> // Based on the configured map output class, set the correct reducer to properly<a name="line.595"></a> -<span class="sourceLineNo">596</span> // sort the incoming values.<a name="line.596"></a> -<span class="sourceLineNo">597</span> // TODO it would be nice to pick one or the other of these formats.<a name="line.597"></a> -<span class="sourceLineNo">598</span> if (KeyValue.class.equals(job.getMapOutputValueClass())) {<a name="line.598"></a> -<span class="sourceLineNo">599</span> job.setReducerClass(KeyValueSortReducer.class);<a name="line.599"></a> -<span class="sourceLineNo">600</span> } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.600"></a> -<span class="sourceLineNo">601</span> job.setReducerClass(PutSortReducer.class);<a name="line.601"></a> -<span class="sourceLineNo">602</span> } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.602"></a> -<span class="sourceLineNo">603</span> job.setReducerClass(TextSortReducer.class);<a name="line.603"></a> -<span class="sourceLineNo">604</span> } else {<a name="line.604"></a> -<span class="sourceLineNo">605</span> LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.605"></a> -<span class="sourceLineNo">606</span> }<a name="line.606"></a> -<span class="sourceLineNo">607</span><a name="line.607"></a> -<span class="sourceLineNo">608</span> conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.608"></a> -<span class="sourceLineNo">609</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.609"></a> -<span class="sourceLineNo">610</span> KeyValueSerialization.class.getName());<a name="line.610"></a> -<span class="sourceLineNo">611</span><a name="line.611"></a> -<span class="sourceLineNo">612</span> if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.612"></a> -<span class="sourceLineNo">613</span> LOG.info("bulkload locality sensitive enabled");<a name="line.613"></a> -<span class="sourceLineNo">614</span> }<a name="line.614"></a> -<span class="sourceLineNo">615</span><a name="line.615"></a> -<span class="sourceLineNo">616</span> /* Now get the region start keys for every table required */<a name="line.616"></a> -<span class="sourceLineNo">617</span> List<String> allTableNames = new ArrayList<>(multiTableInfo.size());<a name="line.617"></a> -<span class="sourceLineNo">618</span> List<RegionLocator> regionLocators = new ArrayList<>( multiTableInfo.size());<a name="line.618"></a> -<span class="sourceLineNo">619</span> List<TableDescriptor> tableDescriptors = new ArrayList<>( multiTableInfo.size());<a name="line.619"></a> -<span class="sourceLineNo">620</span><a name="line.620"></a> -<span class="sourceLineNo">621</span> for( TableInfo tableInfo : multiTableInfo )<a name="line.621"></a> -<span class="sourceLineNo">622</span> {<a name="line.622"></a> -<span class="sourceLineNo">623</span> regionLocators.add(tableInfo.getRegionLocator());<a name="line.623"></a> -<span class="sourceLineNo">624</span> allTableNames.add(tableInfo.getRegionLocator().getName().getNameAsString());<a name="line.624"></a> -<span class="sourceLineNo">625</span> tableDescriptors.add(tableInfo.getTableDescriptor());<a name="line.625"></a> -<span class="sourceLineNo">626</span> }<a name="line.626"></a> -<span class="sourceLineNo">627</span> // Record tablenames for creating writer by favored nodes, and decoding compression, block size and other attributes of columnfamily per table<a name="line.627"></a> -<span class="sourceLineNo">628</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, StringUtils.join(allTableNames, Bytes<a name="line.628"></a> -<span class="sourceLineNo">629</span> .toString(tableSeparator)));<a name="line.629"></a> -<span class="sourceLineNo">630</span> List<ImmutableBytesWritable> startKeys = getRegionStartKeys(regionLocators, writeMultipleTables);<a name="line.630"></a> -<span class="sourceLineNo">631</span> // Use table's region boundaries for TOP split points.<a name="line.631"></a> -<span class="sourceLineNo">632</span> LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.632"></a> -<span class="sourceLineNo">633</span> "to match current region count for all tables");<a name="line.633"></a> -<span class="sourceLineNo">634</span> job.setNumReduceTasks(startKeys.size());<a name="line.634"></a> -<span class="sourceLineNo">635</span><a name="line.635"></a> -<span class="sourceLineNo">636</span> configurePartitioner(job, startKeys, writeMultipleTables);<a name="line.636"></a> -<span class="sourceLineNo">637</span> // Set compression algorithms based on column families<a name="line.637"></a> -<span class="sourceLineNo">638</span><a name="line.638"></a> -<span class="sourceLineNo">639</span> conf.set(COMPRESSION_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(compressionDetails,<a name="line.639"></a> -<span class="sourceLineNo">640</span> tableDescriptors));<a name="line.640"></a> -<span class="sourceLineNo">641</span> conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(blockSizeDetails,<a name="line.641"></a> +<span class="sourceLineNo">581</span> static void configureIncrementalLoad(Job job, List<TableInfo> multiTableInfo,<a name="line.581"></a> +<span class="sourceLineNo">582</span> Class<? extends OutputFormat<?, ?>> cls) throws IOException {<a name="line.582"></a> +<span class="sourceLineNo">583</span> Configuration conf = job.getConfiguration();<a name="line.583"></a> +<span class="sourceLineNo">584</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.584"></a> +<span class="sourceLineNo">585</span> job.setOutputValueClass(MapReduceCell.class);<a name="line.585"></a> +<span class="sourceLineNo">586</span> job.setOutputFormatClass(cls);<a name="line.586"></a> +<span class="sourceLineNo">587</span><a name="line.587"></a> +<span class="sourceLineNo">588</span> if (multiTableInfo.stream().distinct().count() != multiTableInfo.size()) {<a name="line.588"></a> +<span class="sourceLineNo">589</span> throw new IllegalArgumentException("Duplicate entries found in TableInfo argument");<a name="line.589"></a> +<span class="sourceLineNo">590</span> }<a name="line.590"></a> +<span class="sourceLineNo">591</span> boolean writeMultipleTables = false;<a name="line.591"></a> +<span class="sourceLineNo">592</span> if (MultiTableHFileOutputFormat.class.equals(cls)) {<a name="line.592"></a> +<span class="sourceLineNo">593</span> writeMultipleTables = true;<a name="line.593"></a> +<span class="sourceLineNo">594</span> conf.setBoolean(MULTI_TABLE_HFILEOUTPUTFORMAT_CONF_KEY, true);<a name="line.594"></a> +<span class="sourceLineNo">595</span> }<a name="line.595"></a> +<span class="sourceLineNo">596</span> // Based on the configured map output class, set the correct reducer to properly<a name="line.596"></a> +<span class="sourceLineNo">597</span> // sort the incoming values.<a name="line.597"></a> +<span class="sourceLineNo">598</span> // TODO it would be nice to pick one or the other of these formats.<a name="line.598"></a> +<span class="sourceLineNo">599</span> if (KeyValue.class.equals(job.getMapOutputValueClass())<a name="line.599"></a> +<span class="sourceLineNo">600</span> || MapReduceCell.class.equals(job.getMapOutputValueClass())) {<a name="line.600"></a> +<span class="sourceLineNo">601</span> job.setReducerClass(CellSortReducer.class);<a name="line.601"></a> +<span class="sourceLineNo">602</span> } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.602"></a> +<span class="sourceLineNo">603</span> job.setReducerClass(PutSortReducer.class);<a name="line.603"></a> +<span class="sourceLineNo">604</span> } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.604"></a> +<span class="sourceLineNo">605</span> job.setReducerClass(TextSortReducer.class);<a name="line.605"></a> +<span class="sourceLineNo">606</span> } else {<a name="line.606"></a> +<span class="sourceLineNo">607</span> LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.607"></a> +<span class="sourceLineNo">608</span> }<a name="line.608"></a> +<span class="sourceLineNo">609</span><a name="line.609"></a> +<span class="sourceLineNo">610</span> conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.610"></a> +<span class="sourceLineNo">611</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.611"></a> +<span class="sourceLineNo">612</span> CellSerialization.class.getName());<a name="line.612"></a> +<span class="sourceLineNo">613</span><a name="line.613"></a> +<span class="sourceLineNo">614</span> if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.614"></a> +<span class="sourceLineNo">615</span> LOG.info("bulkload locality sensitive enabled");<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> /* Now get the region start keys for every table required */<a name="line.618"></a> +<span class="sourceLineNo">619</span> List<String> allTableNames = new ArrayList<>(multiTableInfo.size());<a name="line.619"></a> +<span class="sourceLineNo">620</span> List<RegionLocator> regionLocators = new ArrayList<>( multiTableInfo.size());<a name="line.620"></a> +<span class="sourceLineNo">621</span> List<TableDescriptor> tableDescriptors = new ArrayList<>( multiTableInfo.size());<a name="line.621"></a> +<span class="sourceLineNo">622</span><a name="line.622"></a> +<span class="sourceLineNo">623</span> for( TableInfo tableInfo : multiTableInfo )<a name="line.623"></a> +<span class="sourceLineNo">624</span> {<a name="line.624"></a> +<span class="sourceLineNo">625</span> regionLocators.add(tableInfo.getRegionLocator());<a name="line.625"></a> +<span class="sourceLineNo">626</span> allTableNames.add(tableInfo.getRegionLocator().getName().getNameAsString());<a name="line.626"></a> +<span class="sourceLineNo">627</span> tableDescriptors.add(tableInfo.getTableDescriptor());<a name="line.627"></a> +<span class="sourceLineNo">628</span> }<a name="line.628"></a> +<span class="sourceLineNo">629</span> // Record tablenames for creating writer by favored nodes, and decoding compression, block size and other attributes of columnfamily per table<a name="line.629"></a> +<span class="sourceLineNo">630</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, StringUtils.join(allTableNames, Bytes<a name="line.630"></a> +<span class="sourceLineNo">631</span> .toString(tableSeparator)));<a name="line.631"></a> +<span class="sourceLineNo">632</span> List<ImmutableBytesWritable> startKeys = getRegionStartKeys(regionLocators, writeMultipleTables);<a name="line.632"></a> +<span class="sourceLineNo">633</span> // Use table's region boundaries for TOP split points.<a name="line.633"></a> +<span class="sourceLineNo">634</span> LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.634"></a> +<span class="sourceLineNo">635</span> "to match current region count for all tables");<a name="line.635"></a> +<span class="sourceLineNo">636</span> job.setNumReduceTasks(startKeys.size());<a name="line.636"></a> +<span class="sourceLineNo">637</span><a name="line.637"></a> +<span class="sourceLineNo">638</span> configurePartitioner(job, startKeys, writeMultipleTables);<a name="line.638"></a> +<span class="sourceLineNo">639</span> // Set compression algorithms based on column families<a name="line.639"></a> +<span class="sourceLineNo">640</span><a name="line.640"></a> +<span class="sourceLineNo">641</span> conf.set(COMPRESSION_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(compressionDetails,<a name="line.641"></a> <span class="sourceLineNo">642</span> tableDescriptors));<a name="line.642"></a> -<span class="sourceLineNo">643</span> conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(bloomTypeDetails,<a name="line.643"></a> +<span class="sourceLineNo">643</span> conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(blockSizeDetails,<a name="line.643"></a> <span class="sourceLineNo">644</span> tableDescriptors));<a name="line.644"></a> -<span class="sourceLineNo">645</span> conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.645"></a> -<span class="sourceLineNo">646</span> serializeColumnFamilyAttribute(dataBlockEncodingDetails, tableDescriptors));<a name="line.646"></a> -<span class="sourceLineNo">647</span><a name="line.647"></a> -<span class="sourceLineNo">648</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.648"></a> -<span class="sourceLineNo">649</span> TableMapReduceUtil.initCredentials(job);<a name="line.649"></a> -<span class="sourceLineNo">650</span> LOG.info("Incremental output configured for tables: " + StringUtils.join(allTableNames, ","));<a name="line.650"></a> -<span class="sourceLineNo">651</span> }<a name="line.651"></a> -<span class="sourceLineNo">652</span><a name="line.652"></a> -<span class="sourceLineNo">653</span> public static void configureIncrementalLoadMap(Job job, TableDescriptor tableDescriptor) throws<a name="line.653"></a> -<span class="sourceLineNo">654</span> IOException {<a name="line.654"></a> -<span class="sourceLineNo">655</span> Configuration conf = job.getConfiguration();<a name="line.655"></a> -<span class="sourceLineNo">656</span><a name="line.656"></a> -<span class="sourceLineNo">657</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.657"></a> -<span class="sourceLineNo">658</span> job.setOutputValueClass(KeyValue.class);<a name="line.658"></a> -<span class="sourceLineNo">659</span> job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.659"></a> -<span class="sourceLineNo">660</span><a name="line.660"></a> -<span class="sourceLineNo">661</span> ArrayList<TableDescriptor> singleTableDescriptor = new ArrayList<>(1);<a name="line.661"></a> -<span class="sourceLineNo">662</span> singleTableDescriptor.add(tableDescriptor);<a name="line.662"></a> -<span class="sourceLineNo">663</span><a name="line.663"></a> -<span class="sourceLineNo">664</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, tableDescriptor.getTableName().getNameAsString());<a name="line.664"></a> -<span class="sourceLineNo">665</span> // Set compression algorithms based on column families<a name="line.665"></a> -<span class="sourceLineNo">666</span> conf.set(COMPRESSION_FAMILIES_CONF_KEY,<a name="line.666"></a> -<span class="sourceLineNo">667</span> serializeColumnFamilyAttribute(compressionDetails, singleTableDescriptor));<a name="line.667"></a> -<span class="sourceLineNo">668</span> conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY,<a name="line.668"></a> -<span class="sourceLineNo">669</span> serializeColumnFamilyAttribute(blockSizeDetails, singleTableDescriptor));<a name="line.669"></a> -<span class="sourceLineNo">670</span> conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY,<a name="line.670"></a> -<span class="sourceLineNo">671</span> serializeColumnFamilyAttribute(bloomTypeDetails, singleTableDescriptor));<a name="line.671"></a> -<span class="sourceLineNo">672</span> conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.672"></a> -<span class="sourceLineNo">673</span> serializeColumnFamilyAttribute(dataBlockEncodingDetails, singleTableDescriptor));<a name="line.673"></a> -<span class="sourceLineNo">674</span><a name="line.674"></a> -<span class="sourceLineNo">675</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.675"></a> -<span class="sourceLineNo">676</span> TableMapReduceUtil.initCredentials(job);<a name="line.676"></a> -<span class="sourceLineNo">677</span> LOG.info("Incremental table " + tableDescriptor.getTableName() + " output configured.");<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> /**<a name="line.680"></a> -<span class="sourceLineNo">681</span> * Runs inside the task to deserialize column family to compression algorithm<a name="line.681"></a> -<span class="sourceLineNo">682</span> * map from the configuration.<a name="line.682"></a> -<span class="sourceLineNo">683</span> *<a name="line.683"></a> -<span class="sourceLineNo">684</span> * @param conf to read the serialized values from<a name="line.684"></a> -<span class="sourceLineNo">685</span> * @return a map from column family to the configured compression algorithm<a name="line.685"></a> -<span class="sourceLineNo">686</span> */<a name="line.686"></a> -<span class="sourceLineNo">687</span> @VisibleForTesting<a name="line.687"></a> -<span class="sourceLineNo">688</span> static Map<byte[], Algorithm> createFamilyCompressionMap(Configuration<a name="line.688"></a> -<span class="sourceLineNo">689</span> conf) {<a name="line.689"></a> -<span class="sourceLineNo">690</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.690"></a> -<span class="sourceLineNo">691</span> COMPRESSION_FAMILIES_CONF_KEY);<a name="line.691"></a> -<span class="sourceLineNo">692</span> Map<byte[], Algorithm> compressionMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.692"></a> -<span class="sourceLineNo">693</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.693"></a> -<span class="sourceLineNo">694</span> Algorithm algorithm = HFileWriterImpl.compressionByName(e.getValue());<a name="line.694"></a> -<span class="sourceLineNo">695</span> compressionMap.put(e.getKey(), algorithm);<a name="line.695"></a> -<span class="sourceLineNo">696</span> }<a name="line.696"></a> -<span class="sourceLineNo">697</span> return compressionMap;<a name="line.697"></a> -<span class="sourceLineNo">698</span> }<a name="line.698"></a> -<span class="sourceLineNo">699</span><a name="line.699"></a> -<span class="sourceLineNo">700</span> /**<a name="line.700"></a> -<span class="sourceLineNo">701</span> * Runs inside the task to deserialize column family to bloom filter type<a name="line.701"></a> -<span class="sourceLineNo">702</span> * map from the configuration.<a name="line.702"></a> -<span class="sourceLineNo">703</span> *<a name="line.703"></a> -<span class="sourceLineNo">704</span> * @param conf to read the serialized values from<a name="line.704"></a> -<span class="sourceLineNo">705</span> * @return a map from column family to the the configured bloom filter type<a name="line.705"></a> -<span class="sourceLineNo">706</span> */<a name="line.706"></a> -<span class="sourceLineNo">707</span> @VisibleForTesting<a name="line.707"></a> -<span class="sourceLineNo">708</span> static Map<byte[], BloomType> createFamilyBloomTypeMap(Configuration conf) {<a name="line.708"></a> -<span class="sourceLineNo">709</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.709"></a> -<span class="sourceLineNo">710</span> BLOOM_TYPE_FAMILIES_CONF_KEY);<a name="line.710"></a> -<span class="sourceLineNo">711</span> Map<byte[], BloomType> bloomTypeMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.711"></a> -<span class="sourceLineNo">712</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.712"></a> -<span class="sourceLineNo">713</span> BloomType bloomType = BloomType.valueOf(e.getValue());<a name="line.713"></a> -<span class="sourceLineNo">714</span> bloomTypeMap.put(e.getKey(), bloomType);<a name="line.714"></a> -<span class="sourceLineNo">715</span> }<a name="line.715"></a> -<span class="sourceLineNo">716</span> return bloomTypeMap;<a name="line.716"></a> -<span class="sourceLineNo">717</span> }<a name="line.717"></a> -<span class="sourceLineNo">718</span><a name="line.718"></a> -<span class="sourceLineNo">719</span> /**<a name="line.719"></a> -<span class="sourceLineNo">720</span> * Runs inside the task to deserialize column family to block size<a name="line.720"></a> -<span class="sourceLineNo">721</span> * map from the configuration.<a name="line.721"></a> -<span class="sourceLineNo">722</span> *<a name="line.722"></a> -<span class="sourceLineNo">723</span> * @param conf to read the serialized values from<a name="line.723"></a> -<span class="sourceLineNo">724</span> * @return a map from column family to the configured block size<a name="line.724"></a> -<span class="sourceLineNo">725</span> */<a name="line.725"></a> -<span class="sourceLineNo">726</span> @VisibleForTesting<a name="line.726"></a> -<span class="sourceLineNo">727</span> static Map<byte[], Integer> createFamilyBlockSizeMap(Configuration conf) {<a name="line.727"></a> -<span class="sourceLineNo">728</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.728"></a> -<span class="sourceLineNo">729</span> BLOCK_SIZE_FAMILIES_CONF_KEY);<a name="line.729"></a> -<span class="sourceLineNo">730</span> Map<byte[], Integer> blockSizeMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.730"></a> -<span class="sourceLineNo">731</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.731"></a> -<span class="sourceLineNo">732</span> Integer blockSize = Integer.parseInt(e.getValue());<a name="line.732"></a> -<span class="sourceLineNo">733</span> blockSizeMap.put(e.getKey(), blockSize);<a name="line.733"></a> -<span class="sourceLineNo">734</span> }<a name="line.734"></a> -<span class="sourceLineNo">735</span> return blockSizeMap;<a name="line.735"></a> -<span class="sourceLineNo">736</span> }<a name="line.736"></a> -<span class="sourceLineNo">737</span><a name="line.737"></a> -<span class="sourceLineNo">738</span> /**<a name="line.738"></a> -<span class="sourceLineNo">739</span> * Runs inside the task to deserialize column family to data block encoding<a name="line.739"></a> -<span class="sourceLineNo">740</span> * type map from the configuration.<a name="line.740"></a> -<span class="sourceLineNo">741</span> *<a name="line.741"></a> -<span class="sourceLineNo">742</span> * @param conf to read the serialized values from<a name="line.742"></a> -<span class="sourceLineNo">743</span> * @return a map from column family to HFileDataBlockEncoder for the<a name="line.743"></a> -<span class="sourceLineNo">744</span> * configured data block type for the family<a name="line.744"></a> -<span class="sourceLineNo">745</span> */<a name="line.745"></a> -<span class="sourceLineNo">746</span> @VisibleForTesting<a name="line.746"></a> -<span class="sourceLineNo">747</span> static Map<byte[], DataBlockEncoding> createFamilyDataBlockEncodingMap(<a name="line.747"></a> -<span class="sourceLineNo">748</span> Configuration conf) {<a name="line.748"></a> -<span class="sourceLineNo">749</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.749"></a> -<span class="sourceLineNo">750</span> DATABLOCK_ENCODING_FAMILIES_CONF_KEY);<a name="line.750"></a> -<span class="sourceLineNo">751</span> Map<byte[], DataBlockEncoding> encoderMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.751"></a> -<span class="sourceLineNo">752</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.752"></a> -<span class="sourceLineNo">753</span> encoderMap.put(e.getKey(), DataBlockEncoding.valueOf((e.getValue())));<a name="line.753"></a> -<span class="sourceLineNo">754</span> }<a name="line.754"></a> -<span class="sourceLineNo">755</span> return encoderMap;<a name="line.755"></a> -<span class="sourceLineNo">756</span> }<a name="line.756"></a> -<span class="sourceLineNo">757</span><a name="line.757"></a> -<span class="sourceLineNo">758</span><a name="line.758"></a> -<span class="sourceLineNo">759</span> /**<a name="line.759"></a> -<span class="sourceLineNo">760</span> * Run inside the task to deserialize column family to given conf value map.<a name="line.760"></a> -<span class="sourceLineNo">761</span> *<a name="line.761"></a> -<span class="sourceLineNo">762</span> * @param conf to read the serialized values from<a name="line.762"></a> -<span class="sourceLineNo">763</span> * @param confName conf key to read from the configuration<a name="line.763"></a> -<span class="sourceLineNo">764</span> * @return a map of column family to the given configuration value<a name="line.764"></a> -<span class="sourceLineNo">765</span> */<a name="line.765"></a> -<span class="sourceLineNo">766</span> private static Map<byte[], String> createFamilyConfValueMap(<a name="line.766"></a> -<span class="sourceLineNo">767</span> Configuration conf, String confName) {<a name="line.767"></a> -<span class="sourceLineNo">768</span> Map<byte[], String> confValMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.768"></a> -<span class="sourceLineNo">769</span> String confVal = conf.get(confName, "");<a name="line.769"></a> -<span class="sourceLineNo">770</span> for (String familyConf : confVal.split("&")) {<a name="line.770"></a> -<span class="sourceLineNo">771</span> String[] familySplit = familyConf.split("=");<a name="line.771"></a> -<span class="sourceLineNo">772</span> if (familySplit.length != 2) {<a name="line.772"></a> -<span class="sourceLineNo">773</span> continue;<a name="line.773"></a> -<span class="sourceLineNo">774</span> }<a name="line.774"></a> -<span class="sourceLineNo">775</span> try {<a name="line.775"></a> -<span class="sourceLineNo">776</span> confValMap.put(URLDecoder.decode(familySplit[0], "UTF-8").getBytes(StandardCharsets.UTF_8),<a name="line.776"></a> -<span class="sourceLineNo">777</span> URLDecoder.decode(familySplit[1], "UTF-8"));<a name="line.777"></a> -<span class="sourceLineNo">778</span> } catch (UnsupportedEncodingException e) {<a name="line.778"></a> -<span class="sourceLineNo">779</span> // will not happen with UTF-8 encoding<a name="line.779"></a> -<span class="sourceLineNo">780</span> throw new AssertionError(e);<a name="line.780"></a> -<span class="sourceLineNo">781</span> }<a name="line.781"></a> -<span class="sourceLineNo">782</span> }<a name="line.782"></a> -<span class="sourceLineNo">783</span> return confValMap;<a name="line.783"></a> -<span class="sourceLineNo">784</span> }<a name="line.784"></a> -<span class="sourceLineNo">785</span><a name="line.785"></a> -<span class="sourceLineNo">786</span> /**<a name="line.786"></a> -<span class="sourceLineNo">787</span> * Configure <code>job</code> with a TotalOrderPartitioner, partitioning against<a name="line.787"></a> -<span class="sourceLineNo">788</span> * <code>splitPoints</code>. Cleans up the partitions file after job exists.<a name="line.788"></a> -<span class="sourceLineNo">789</span> */<a name="line.789"></a> -<span class="sourceLineNo">790</span> static void configurePartitioner(Job job, List<ImmutableBytesWritable> splitPoints, boolean<a name="line.790"></a> -<span class="sourceLineNo">791</span> writeMultipleTables)<a name="line.791"></a> -<span class="sourceLineNo">792</span> throws IOException {<a name="line.792"></a> -<span class="sourceLineNo">793</span> Configuration conf = job.getConfiguration();<a name="line.793"></a> -<span class="sourceLineNo">794</span> // create the partitions file<a name="line.794"></a> -<span class="sourceLineNo">795</span> FileSystem fs = FileSystem.get(conf);<a name="line.795"></a> -<span class="sourceLineNo">796</span> String hbaseTmpFsDir =<a name="line.796"></a> -<span class="sourceLineNo">797</span> conf.get(HConstants.TEMPORARY_FS_DIRECTORY_KEY,<a name="line.797"></a> -<span class="sourceLineNo">798</span> HConstants.DEFAULT_TEMPORARY_HDFS_DIRECTORY);<a name="line.798"></a> -<span class="sourceLineNo">799</span> Path partitionsPath = new Path(hbaseTmpFsDir, "partitions_" + UUID.randomUUID());<a name="line.799"></a> -<span class="sourceLineNo">800</span> fs.makeQualified(partitionsPath);<a name="line.800"></a> -<span class="sourceLineNo">801</span> writePartitions(conf, partitionsPath, splitPoints, writeMultipleTables);<a name="line.801"></a> -<span class="sourceLineNo">802</span> fs.deleteOnExit(partitionsPath);<a name="line.802"></a> -<span class="sourceLineNo">803</span><a name="line.803"></a> -<span class="sourceLineNo">804</span> // configure job to use it<a name="line.804"></a> -<span class="sourceLineNo">805</span> job.setPartitionerClass(TotalOrderPartitioner.class);<a name="line.805"></a> -<span class="sourceLineNo">806</span> TotalOrderPartitioner.setPartitionFile(conf, partitionsPath);<a name="line.806"></a> -<span class="sourceLineNo">807</span> }<a name="line.807"></a> -<span class="sourceLineNo">808</span><a name="line.808"></a> -<span class="sourceLineNo">809</span> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "RCN_REDUNDANT_NULLCHECK_OF_NONNULL_VALUE")<a name="line.809"></a> -<span class="sourceLineNo">810</span> @VisibleForTesting<a name="line.810"></a> -<span class="sourceLineNo">811</span> static String serializeColumnFamilyAttribute(Function<ColumnFamilyDescriptor, String> fn, List<TableDescriptor> allTables)<a name="line.811"></a> -<span class="sourceLineNo">812</span> throws UnsupportedEncodingException {<a name="line.812"></a> -<span class="sourceLineNo">813</span> StringBuilder attributeValue = new StringBuilder();<a name="line.813"></a> -<span class="sourceLineNo">814</span> int i = 0;<a name="line.814"></a> -<span class="sourceLineNo">815</span> for (TableDescriptor tableDescriptor : allTables) {<a name="line.815"></a> -<span class="sourceLineNo">816</span> if (tableDescriptor == null) {<a name="line.816"></a> -<span class="sourceLineNo">817</span> // could happen with mock table instance<a name="line.817"></a> -<span class="sourceLineNo">818</span> // CODEREVIEW: Can I set an empty string in conf if mock table instance?<a name="line.818"></a> -<span class="sourceLineNo">819</span> return "";<a name="line.819"></a> -<span class="sourceLineNo">820</span> }<a name="line.820"></a> -<span class="sourceLineNo">821</span> for (ColumnFamilyDescriptor familyDescriptor : tableDescriptor.getColumnFamilies()) {<a name="line.821"></a> -<span class="sourceLineNo">822</span> if (i++ > 0) {<a name="line.822"></a> -<span class="sourceLineNo">823</span> attributeValue.append('&');<a name="line.823"></a> -<span class="sourceLineNo">824</span> }<a name="line.824"></a> -<span class="sourceLineNo">825</span> attributeValue.append(URLEncoder.encode(<a name="line.825"></a> -<span class="sourceLineNo">826</span> Bytes.toString(combineTableNameSuffix(tableDescriptor.getTableName().getName(), familyDescriptor.getName())),<a name="line.826"></a> -<span class="sourceLineNo">827</span> "UTF-8"));<a name="line.827"></a> -<span class="sourceLineNo">828</span> attributeValue.append('=');<a name="line.828"></a> -<span class="sourceLineNo">829</span> attributeValue.append(URLEncoder.encode(fn.apply(familyDescriptor), "UTF-8"));<a name="line.829"></a> -<span class="sourceLineNo">830</span> }<a name="line.830"></a> -<span class="sourceLineNo">831</span> }<a name="line.831"></a> -<span class="sourceLineNo">832</span> // Get rid of the last ampersand<a name="line.832"></a> -<span class="sourceLineNo">833</span> return attributeValue.toString();<a name="line.833"></a> -<span class="sourceLineNo">834</span> }<a name="line.834"></a> -<span class="sourceLineNo">835</span><a name="line.835"></a> -<span class="sourceLineNo">836</span> /**<a name="line.836"></a> -<span class="sourceLineNo">837</span> * Serialize column family to compression algorithm map to configuration.<a name="line.837"></a> -<span class="sourceLineNo">838</span> * Invoked while configuring the MR job for incremental load.<a name="line.838"></a> -<span class="sourceLineNo">839</span> *<a name="line.839"></a> -<span class="sourceLineNo">840</span> * @param tableDescriptor to read the properties from<a name="line.840"></a> -<span class="sourceLineNo">841</span> * @param conf to persist serialized values into<a name="line.841"></a> -<span class="sourceLineNo">842</span> * @throws IOException<a name="line.842"></a> -<span class="sourceLineNo">843</span> * on failure to read column family descriptors<a name="line.843"></a> -<span class="sourceLineNo">844</span> */<a name="line.844"></a> -<span class="sourceLineNo">845</span> @VisibleForTesting<a name="line.845"></a> -<span class="sourceLineNo">846</span> static Function<ColumnFamilyDescriptor, String> compressionDetails = familyDescriptor -><a name="line.846"></a> -<span class="sourceLineNo">847</span> familyDescriptor.getCompressionType().getName();<a name="line.847"></a> -<span class="sourceLineNo">848</span><a name="line.848"></a> -<span class="sourceLineNo">849</span> /**<a name="line.849"></a> -<span class="sourceLineNo">850</span> * Serialize column family to block size map to configuration. Invoked while<a name="line.850"></a> -<span class="sourceLineNo">851</span> * configuring the MR job for incremental load.<a name="line.851"></a> -<span class="sourceLineNo">852</span> *<a name="line.852"></a> -<span class="sourceLineNo">853</span> * @param tableDescriptor<a name="line.853"></a> -<span class="sourceLineNo">854</span> * to read the properties from<a name="line.854"></a> -<span class="sourceLineNo">855</span> * @param conf<a name="line.855"></a> -<span class="sourceLineNo">856</span> * to persist serialized values into<a name="line.856"></a> -<span class="sourceLineNo">857</span> *<a name="line.857"></a> -<span class="sourceLineNo">858</span> * @throws IOException<a name="line.858"></a> -<span class="sourceLineNo">859</span> * on failure to read column family descriptors<a name="line.859"></a> -<span class="sourceLineNo">860</span> */<a name="line.860"></a> -<span class="sourceLineNo">861</span> @VisibleForTesting<a name="line.861"></a> -<span class="sourceLineNo">862</span> static Function<ColumnFamilyDescriptor, String> blockSizeDetails = familyDescriptor -> String<a name="line.862"></a> -<span class="sourceLineNo">863</span> .valueOf(familyDescriptor.getBlocksize());<a name="line.863"></a> -<span class="sourceLineNo">864</span><a name="line.864"></a> -<span class="sourceLineNo">865</span> /**<a name="line.865"></a> -<span class="sourceLineNo">866</span> * Serialize column family to bloom type map to configuration. Invoked while<a name="line.866"></a> -<span class="sourceLineNo">867</span> * configuring the MR job for incremental load.<a name="line.867"></a> -<span class="sourceLineNo">868</span> *<a name="line.868"></a> -<span class="sourceLineNo">869</span> * @param tableDescriptor<a name="line.869"></a> -<span class="sourceLineNo">870</span> * to read the properties from<a name="line.870"></a> -<span class="sourceLineNo">871</span> * @param conf<a name="line.871"></a> -<span class="sourceLineNo">872</span> * to persist serialized values into<a name="line.872"></a> -<span class="sourceLineNo">873</span> *<a name="line.873"></a> -<span class="sourceLineNo">874</span> * @throws IOException<a name="line.874"></a> -<span class="sourceLineNo">875</span> * on failure to read column family descriptors<a name="line.875"></a> -<span class="sourceLineNo">876</span> */<a name="line.876"></a> -<span class="sourceLineNo">877</span> @VisibleForTesting<a name="line.877"></a> -<span class="sourceLineNo">878</span> static Function<ColumnFamilyDescriptor, String> bloomTypeDetails = familyDescriptor -> {<a name="line.878"></a> -<span class="sourceLineNo">879</span> String bloomType = familyDescriptor.getBloomFilterType().toString();<a name="line.879"></a> -<span class="sourceLineNo">880</span> if (bloomType == null) {<a name="line.880"></a> -<span class="sourceLineNo">881</span> bloomType = ColumnFamilyDescriptorBuilder.DEFAULT_BLOOMFILTER.name();<a name="line.881"></a> -<span class="sourceLineNo">882</span> }<a name="line.882"></a> -<span class="sourceLineNo">883</span> return bloomType;<a name="line.883"></a> -<span class="sourceLineNo">884</span> };<a name="line.884"></a> -<span class="sourceLineNo">885</span><a name="line.885"></a> -<span class="sourceLineNo">886</span> /**<a name="line.886"></a> -<span class="sourceLineNo">887</span> * Serialize column family to data block encoding map to configuration.<a name="line.887"></a> -<span class="sourceLineNo">888</span> * Invoked while configuring the MR job for incremental load.<a name="line.888"></a> -<span class="sourceLineNo">889</span> *<a name="line.889"></a> -<span class="sourceLineNo">890</span> * @param tableDescriptor<a name="line.890"></a> -<span class="sourceLineNo">891</span> * to read the properties from<a name="line.891"></a> -<span class="sourceLineNo">892</span> * @param conf<a name="line.892"></a> -<span class="sourceLineNo">893</span> * to persist serialized values into<a name="line.893"></a> -<span class="sourceLineNo">894</span> * @throws IOException<a name="line.894"></a> -<span class="sourceLineNo">895</span> * on failure to read column family descriptors<a name="line.895"></a> -<span class="sourceLineNo">896</span> */<a name="line.896"></a> -<span class="sourceLineNo">897</span> @VisibleForTesting<a name="line.897"></a> -<span class="sourceLineNo">898</span> static Function<ColumnFamilyDescriptor, String> dataBlockEncodingDetails = familyDescriptor -> {<a name="line.898"></a> -<span class="sourceLineNo">899</span> DataBlockEncoding encoding = familyDescriptor.getDataBlockEncoding();<a name="line.899"></a> -<span class="sourceLineNo">900</span> if (encoding == null) {<a name="line.900"></a> -<span class="sourceLineNo">901</span> encoding = DataBlockEncoding.NONE;<a name="line.901"></a> -<span class="sourceLineNo">902</span> }<a name="line.902"></a> -<span class="sourceLineNo">903</span> return encoding.toString();<a name="line.903"></a> -<span class="sourceLineNo">904</span> };<a name="line.904"></a> -<span class="sourceLineNo">905</span><a name="line.905"></a> -<span class="sourceLineNo">906</span>}<a name="line.906"></a> +<span class="sourceLineNo">645</span> conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(bloomTypeDetails,<a name="line.645"></a> +<span class="sourceLineNo">646</span> tableDescriptors));<a name="line.646"></a> +<span class="sourceLineNo">647</span> conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.647"></a> +<span class="sourceLineNo">648</span> serializeColumnFamilyAttribute(dataBlockEncodingDetails, tableDescriptors));<a name="line.648"></a> +<span class="sourceLineNo">649</span><a name="line.649"></a> +<span class="sourceLineNo">650</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.650"></a> +<span class="sourceLineNo">651</span> TableMapReduceUtil.initCredentials(job);<a name="line.651"></a> +<span class="sourceLineNo">652</span> LOG.info("Incremental output configured for tables: " + StringUtils.join(allTableNames, ","));<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> public static void configureIncrementalLoadMap(Job job, TableDescriptor tableDescriptor) throws<a name="line.655"></a> +<span class="sourceLineNo">656</span> IOException {<a name="line.656"></a> +<span class="sourceLineNo">657</span> Configuration conf = job.getConfiguration();<a name="line.657"></a> +<span class="sourceLineNo">658</span><a name="line.658"></a> +<span class="sourceLineNo">659</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.659"></a> +<span class="sourceLineNo">660</span> job.setOutputValueClass(MapReduceCell.class);<a name="line.660"></a> +<span class="sourceLineNo">661</span> job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.661"></a> +<span class="sourceLineNo">662</span><a name="line.662"></a> +<span class="sourceLineNo">663</span> ArrayList<TableDescriptor> singleTableDescriptor = new ArrayList<>(1);<a name="line.663"></a> +<span class="sourceLineNo">664</span> singleTableDescriptor.add(tableDescriptor);<a name="line.664"></a> +<span class="sourceLineNo">665</span><a name="line.665"></a> +<span class="sourceLineNo">666</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, tableDescriptor.getTableName().getNameAsString());<a name="line.666"></a> +<span class="sourceLineNo">667</span> // Set compression algorithms based on column families<a name="line.667"></a> +<span class="sourceLineNo">668</span> conf.set(COMPRESSION_FAMILIES_CONF_KEY,<a name="line.668"></a> +<span class="sourceLineNo">669</span> serializeColumnFamilyAttribute(compressionDetails, singleTableDescriptor));<a name="line.669"></a> +<span class="sourceLineNo">670</span> conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY,<a name="line.670"></a> +<span class="sourceLineNo">671</span> serializeColumnFamilyAttribute(blockSizeDetails, singleTableDescriptor));<a name="line.671"></a> +<span class="sourceLineNo">672</span> conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY,<a name="line.672"></a> +<span class="sourceLineNo">673</span> serializeColumnFamilyAttribute(bloomTypeDetails, singleTableDescriptor));<a name="line.673"></a> +<span class="sourceLineNo">674</span> conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.674"></a> +<span class="sourceLineNo">675</span> serializeColumnFamilyAttribute(dataBlockEncodingDetails, singleTableDescriptor));<a name="line.675"></a> +<span class="sourceLineNo">676</span><a name="line.676"></a> +<span class="sourceLineNo">677</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.677"></a> +<span class="sourceLineNo">678</span> TableMapReduceUtil.initCredentials(job);<a name="line.678"></a> +<span class="sourceLineNo">679</span> LOG.info("Incremental table " + tableDescriptor.getTableName() + " output configured.");<a name="line.679"></a> +<span class="sourceLineNo">680</span> }<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> * Runs inside the task to deserialize column family to compression algorithm<a name="line.683"></a> +<span class="sourceLineNo">684</span> * map from the configuration.<a name="line.684"></a> +<span class="sourceLineNo">685</span> *<a name="line.685"></a> +<span class="sourceLineNo">686</span> * @param conf to read the serialized values from<a name="line.686"></a> +<span class="sourceLineNo">687</span> * @return a map from column family to the configured compression algorithm<a name="line.687"></a> +<span class="sourceLineNo">688</span> */<a name="line.688"></a> +<span class="sourceLineNo">689</span> @VisibleForTesting<a name="line.689"></a> +<span class="sourceLineNo">690</span> static Map<byte[], Algorithm> createFamilyCompressionMap(Configuration<a name="line.690"></a> +<span class="sourceLineNo">691</span> conf) {<a name="line.691"></a> +<span class="sourceLineNo">692</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.692"></a> +<span class="sourceLineNo">693</span> COMPRESSION_FAMILIES_CONF_KEY);<a name="line.693"></a> +<span class="sourceLineNo">694</span> Map<byte[], Algorithm> compressionMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.694"></a> +<span class="sourceLineNo">695</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.695"></a> +<span class="sourceLineNo">696</span> Algorithm algorithm = HFileWriterImpl.compressionByName(e.getValue());<a name="line.696"></a> +<span class="sourceLineNo">697</span> compressionMap.put(e.getKey(), algorithm);<a name="line.697"></a> +<span class="sourceLineNo">698</span> }<a name="line.698"></a> +<span class="sourceLineNo">699</span> return compressionMap;<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> * Runs inside the task to deserialize column family to bloom filter type<a name="line.703"></a> +<span class="sourceLineNo">704</span> * map from the configuration.<a name="line.704"></a> +<span class="sourceLineNo">705</span> *<a name="line.705"></a> +<span class="sourceLineNo">706</span> * @param conf to read the serialized values from<a name="line.706"></a> +<span class="sourceLineNo">707</span> * @return a map from column family to the the configured bloom filter type<a name="line.707"></a> +<span class="sourceLineNo">708</span> */<a name="line.708"></a> +<span class="sourceLineNo">709</span> @VisibleForTesting<a name="line.709"></a> +<span class="sourceLineNo">710</span> static Map<byte[], BloomType> createFamilyBloomTypeMap(Configuration conf) {<a name="line.710"></a> +<span class="sourceLineNo">711</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.711"></a> +<span class="sourceLineNo">712</span> BLOOM_TYPE_FAMILIES_CONF_KEY);<a name="line.712"></a> +<span class="sourceLineNo">713</span> Map<byte[], BloomType> bloomTypeMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.713"></a> +<span class="sourceLineNo">714</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.714"></a> +<span class="sourceLineNo">715</span> BloomType bloomType = BloomType.valueOf(e.getValue());<a name="line.715"></a> +<span class="sourceLineNo">716</span> bloomTypeMap.put(e.getKey(), bloomType);<a name="line.716"></a> +<span class="sourceLineNo">717</span> }<a name="line.717"></a> +<span class="sourceLineNo">718</span> return bloomTypeMap;<a name="line.718"></a> +<span class="sourceLineNo">719</span> }<a name="line.719"></a> +<span class="sourceLineNo">720</span><a name="line.720"></a> +<span class="sourceLineNo">721</span> /**<a name="line.721"></a> +<span class="sourceLineNo">722</span> * Runs inside the task to deserialize column family to block size<a name="line.722"></a> +<span class="sourceLineNo">723</span> * map from the configuration.<a name="line.723"></a> +<span class="sourceLineNo">724</span> *<a name="line.724"></a> +<span class="sourceLineNo">725</span> * @param conf to read the serialized values from<a name="line.725"></a> +<span class="sourceLineNo">726</span> * @return a map from column family to the configured block size<a name="line.726"></a> +<span class="sourceLineNo">727</span> */<a name="line.727"></a> +<span class="sourceLineNo">728</span> @VisibleForTesting<a name="line.728"></a> +<span class="sourceLineNo">729</span> static Map<byte[], Integer> createFamilyBlockSizeMap(Configuration conf) {<a name="line.729"></a> +<span class="sourceLineNo">730</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.730"></a> +<span class="sourceLineNo">731</span> BLOCK_SIZE_FAMILIES_CONF_KEY);<a name="line.731"></a> +<span class="sourceLineNo">732</span> Map<byte[], Integer> blockSizeMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.732"></a> +<span class="sourceLineNo">733</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.733"></a> +<span class="sourceLineNo">734</span> Integer blockSize = Integer.parseInt(e.getValue());<a name="line.734"></a> +<span class="sourceLineNo">735</span> blockSizeMap.put(e.getKey(), blockSize);<a name="line.735"></a> +<span class="sourceLineNo">736</span> }<a name="line.736"></a> +<span class="sourceLineNo">737</span> return blockSizeMap;<a name="line.737"></a> +<span class="sourceLineNo">738</span> }<a name="line.738"></a> +<span class="sourceLineNo">739</span><a name="line.739"></a> +<span class="sourceLineNo">740</span> /**<a name="line.740"></a> +<span class="sourceLineNo">741</span> * Runs inside the task to deserialize column family to data block encoding<a name="line.741"></a> +<span class="sourceLineNo">742</span> * type map from the configuration.<a name="line.742"></a> +<span class="sourceLineNo">743</span> *<a name="line.743"></a> +<span class="sourceLineNo">744</span> * @param conf to read the serialized values from<a name="line.744"></a> +<span class="sourceLineNo">745</span> * @return a map from column family to HFileDataBlockEncoder for the<a name="line.745"></a> +<span class="sourceLineNo">746</span> * configured data block type for the family<a name="line.746"></a> +<span class="sourceLineNo">747</span> */<a name="line.747"></a> +<span class="sourceLineNo">748</span> @VisibleForTesting<a name="line.748"></a> +<span class="sourceLineNo">749</span> static Map<byte[], DataBlockEncoding> createFamilyDataBlockEncodingMap(<a name="line.749"></a> +<span class="sourceLineNo">750</span> Configuration conf) {<a name="line.750"></a> +<span class="sourceLineNo">751</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.751"></a> +<span class="sourceLineNo">752</span> DATABLOCK_ENCODING_FAMILIES_CONF_KEY);<a name="line.752"></a> +<span class="sourceLineNo">753</span> Map<byte[], DataBlockEncoding> encoderMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.753"></a> +<span class="sourceLineNo">754</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.754"></a> +<span class="sourceLineNo">755</span> encoderMap.put(e.getKey(), DataBlockEncoding.valueOf((e.getValue())));<a name="line.755"></a> +<span class="sourceLineNo">756</span> }<a name="line.756"></a> +<span class="sourceLineNo">757</span> return encoderMap;<a name="line.757"></a> +<span class="sourceLineNo">758</span> }<a name="line.758"></a> +<span class="sourceLineNo">759</span><a name="line.759"></a> +<span class="sourceLineNo">760</span><a name="line.760"></a> +<span class="sourceLineNo">761</span> /**<a name="line.761"></a> +<span class="sourceLineNo">762</span> * Run inside the task to deserialize column family to given conf value map.<a name="line.762"></a> +<span class="sourceLineNo">763</span> *<a name="line.763"></a> +<span class="sourceLineNo">764</span> * @param conf to read the serialized values from<a name="line.764"></a> +<span class="sourceLineNo">765</span> * @param confName conf key to read from the configuration<a name="line.765"></a> +<span class="sourceLineNo">766</span> * @return a map of column family to the given configuration value<a name="line.766"></a> +<span class="sourceLineNo">767</span> */<a name="line.767"></a> +<span class="sourceLineNo">768</span> private static Map<byte[], String> createFamilyConfValueMap(<a name="line.768"></a> +<span class="sourceLineNo">769</span> Configuration conf, String confName) {<a name="line.769"></a> +<span class="sourceLineNo">770</span> Map<byte[], String> confValMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.770"></a> +<span class="sourceLineNo">771</span> String confVal = conf.get(confName, "");<a name="line.771"></a> +<span class="sourceLineNo">772</span> for (String familyConf : confVal.split("&")) {<a name="line.772"></a> +<span class="sourceLineNo">773</span> String[] familySplit = familyConf.split("=");<a name="line.773"></a> +<span class="sourceLineNo">774</span> if (familySplit.length != 2) {<a name="line.774"></a> +<span class="sourceLineNo">775</span> continue;<a name="line.775"></a> +<span class="sourceLineNo">776</span> }<a name="line.776"></a> +<span class="sourceLineNo">777</span> try {<a name="line.777"></a> +<span class="sourceLineNo">778</span> confValMap.put(URLDecoder.decode(familySplit[0], "UTF-8").getBytes(StandardCharsets.UTF_8),<a name="line.778"></a> +<span class="sourceLineNo">779</span> URLDecoder.decode(familySplit[1], "UTF-8"));<a name="line.779"></a> +<span class="sourceLineNo">780</span> } catch (UnsupportedEncodingException e) {<a name="line.780"></a> +<span class="sourceLineNo">781</span> // will not happen with UTF-8 encoding<a name="line.781"></a> +<span class="sourceLineNo">782</span> throw new AssertionError(e);<a name="line.782"></a> +<span class="sourceLineNo">783</span> }<a name="line.783"></a> +<span class="sourceLineNo">784</span> }<a name="line.784"></a> +<span class="sourceLineNo">785</span> return confValMap;<a name="line.785"></a> +<span class="sourceLineNo">786</span> }<a name="line.786"></a> +<span class="sourceLineNo">787</span><a name="line.787"></a> +<span class="sourceLineNo">788</span> /**<a name="line.788"></a> +<span class="sourceLineNo">789</span> * Configure <code>job</code> with a TotalOrderPartitioner, partitioning against<a name="line.789"></a> +<span class="sourceLineNo">790</span> * <code>splitPoints</code>. Cleans up the partitions file after job exists.<a name="line.790"></a> +<span class="sourceLineNo">791</span> */<a name="line.791"></a> +<span class="sourceLineNo">792</span> static void configurePartitioner(Job job, List<ImmutableBytesWritable> splitPoints, boolean<a name="line.792"></a> +<span class="sourceLineNo">793</span> writeMultipleTables)<a name="line.793"></a> +<span class="sourceLineNo">794</span> throws IOException {<a name="line.794"></a> +<span class="sourceLineNo">795</span> Configuration conf = job.getConfiguration();<a name="line.795"></a> +<span class="sourceLineNo">796</span> // create the partitions file<a name="line.796"></a> +<span class="sourceLineNo">797</span> FileSystem fs = FileSystem.get(conf);<a name="line.797"></a> +<span class="sourceLineNo">798</span> String hbaseTmpFsDir =<a name="line.798"></a> +<span class="sourceLineNo">799</span> conf.get(HConstants.TEMPORARY_FS_DIRECTORY_KEY,<a name="line.799"></a> +<span class="sourceLineNo">800</span> HConstants.DEFAULT_TEMPORARY_HDFS_DIRECTORY);<a name="line.800"></a> +<span class="sourceLineNo">801</span> Path partitionsPath = new Path(hbaseTmpFsDir, "partitions_" + UUID.randomUUID());<a name="line.801"></a> +<span class="sourceLineNo">802</span> fs.makeQualified(partitionsPath);<a name="line.802"></a> +<span class="sourceLineNo">803</span> writePartitions(conf, partitionsPath, splitPoints, writeMultipleTables);<a name="line.803"></a> +<span class="sourceLineNo">804</span> fs.deleteOnExit(partitionsPath);<a name="line.804"></a> +<span class="sourceLineNo">805</span><a name="line.805"></a> +<span class="sourceLineNo">806</span> // configure job to use it<a name="line.806"></a> +<span class="sourceLineNo">807</span> job.setPartitionerClass(TotalOrderPartitioner.class);<a name="line.807"></a> +<span class="sourceLineNo">808</span> TotalOrderPartitioner.setPartitionFile(conf, partitionsPath);<a name="line.808"></a> +<span class="sourceLineNo">809</span> }<a name="line.809"></a> +<span class="sourceLineNo">810</span><a name="line.810"></a> +<span class="sourceLineNo">811</span> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "RCN_REDUNDANT_NULLCHECK_OF_NONNULL_VALUE")<a name="line.811"></a> +<span class="sourceLineNo">812</span> @VisibleForTesting<a name="line.812"></a> +<span class="sourceLineNo">813</span> static String serializeColumnFamilyAttribute(Function<ColumnFamilyDescriptor, String> fn, List<TableDescriptor> allTables)<a name="line.813"></a> +<span class="sourceLineNo">814</span> throws UnsupportedEncodingException {<a name="line.814"></a> +<span class="sourceLineNo">815</span> StringBuilder attributeValue = new StringBuilder();<a name="line.815"></a> +<span class="sourceLineNo">816</span> int i = 0;<a name="line.816"></a> +<span class="sourceLineNo">817</span> for (TableDescriptor tableDescriptor : allTables) {<a name="line.817"></a> +<span class="sourceLineNo">818</span> if (tableDescriptor == null) {<a name="line.818"></a> +<span class="sourceLineNo">819</span> // could happen with mock table instance<a name="line.819"></a> +<span class="sourceLineNo">820</span> // CODEREVIEW: Can I set an empty string in conf if mock table instance?<a name="line.820"></a> +<span class="sourceLineNo">821</span> return "";<a name="line.821"></a> +<span class="sourceLineNo">822</span> }<a name="line.822"></a> +<span class="sourceLineNo">823</span> for (ColumnFamilyDescriptor familyDescriptor : tableDescriptor.getColumnFamilies()) {<a name="line.823"></a> +<span class="sourceLineNo">824</span> if (i++ > 0) {<a name="line.824"></a> +<span class="sourceLineNo">825</span> attributeValue.append('&');<a name="line.825"></a> +<span class="sourceLineNo">826</span> }<a name="line.826"></a> +<span class="sourceLineNo">827</span> attributeValue.append(URLEncoder.encode(<a name="line.827"></a> +<span class="sourceLineNo">828</span> Bytes.toString(combineTableNameSuffix(tableDescriptor.getTableName().getName(), familyDescriptor.getName())),<a name="line.828"></a> +<span class="sourceLineNo">829</span> "UTF-8"));<a name="line.829"></a> +<span class="sourceLineNo">830</span> attributeValue.append('=');<a name="line.830"></a> +<span class="sourceLineNo">831</span> attributeValue.append(URLEncoder.encode(fn.apply(familyDescriptor), "UTF-8"));<a name="line.831"></a> +<span class="sourceLineNo">832</span> }<a name="line.832"></a> +<span class="sourceLineNo">833</span> }<a name="line.833"></a> +<span class="sourceLineNo">834</span> // Get rid of the last ampersand<a name="line.834"></a> +<span class="sourceLineNo">835</span> return attributeValue.toString();<a name="line.835"></a> +<span class="sourceLineNo">836</span> }<a name="line.836"></a> +<span class="sourceLineNo">837</span><a name="line.837"></a> +<span class="sourceLineNo">838</span> /**<a name="line.838"></a> +<span class="sourceLineNo">839</span> * Serialize column family to compression algorithm map to configuration.<a name="line.839"></a> +<span class="sourceLineNo">840</span> * Invoked while configuring the MR job for incremental load.<a name="line.840"></a> +<span class="sourceLineNo">841</span> *<a name="line.841"></a> +<span class="sourceLineNo">842</span> * @param tableDescriptor to read the properties from<a name="line.842"></a> +<span class="sourceLineNo">843</span> * @param conf to persist serialized values into<a name="line.843"></a> +<span class="sourceLineNo">844</span> * @throws IOException<a name="line.844"></a> +<span class="sourceLineNo">845</span> * on failure to read column family descriptors<a name="line.845"></a> +<span class="sourceLineNo">846</span> */<a name="line.846"></a> +<span class="sourceLineNo">847</span> @VisibleForTesting<a name="line.847"></a> +<span class="sourceLineNo">848</span> static Function<ColumnFamilyDescriptor, String> compressionDetails = familyDescriptor -><a name="line.848"></a> +<span class="sourceLineNo">849</span> familyDescriptor.getCompressionType().getName();<a name="line.849"></a> +<span class="sourceLineNo">850</span><a name="line.850"></a> +<span class="sourceLineNo">851</span> /**<a name="line.851"></a> +<span class="sourceLineNo">852</span> * Serialize column family to block size map to configuration. Invoked while<a name="line.852"></a> +<span class="sourceLineNo">853</span> * configuring the MR job for incremental load.<a name="line.853"></a> +<span class="sourceLineNo">854</span> *<a name="line.854"></a> +<span class="sourceLineNo">855</span> * @param tableDescriptor<a name="line.855"></a> +<span class="sourceLineNo">856</span> * to read the properties from<a name="line.856"></a> +<span class="sourceLineNo">857</span> * @param conf<a name="line.857"></a> +<span class="sourceLineNo">858</span> * to persist serialized values into<a name="line.858"></a> +<span class="sourceLineNo">859</span> *<a n
<TRUNCATED>
