public static void getData(CloudataConf conf, Path keyPath) throws IOException {
JobConf jobConf = new JobConf(TeraReadJob.class);
jobConf.set("user.name", conf.getUserId());
String libDir = CloudataMapReduceUtil.initMapReduce(jobConf);
Path tempOutputPath = new Path("ManyTableJob_Get_" + System.currentTimeMillis());
jobConf.setJobName("ManyTableJob_Get_" + "(" + new Date() + ")");
TextOutputFormat.setOutputPath(jobConf, tempOutputPath);
// <MAP>
jobConf.setMapperClass(ManyTableGetMap.class);
jobConf.setInputFormat(TextInputFormat.class);
TextInputFormat.addInputPath(jobConf, keyPath);
jobConf.setMapSpeculativeExecution(false);
jobConf.setMaxMapAttempts(0);
// </MAP>
// <REDUCE>
jobConf.setNumReduceTasks(0);
// </REDUCE>
try {
// Run Job
JobClient.runJob(jobConf);
} finally {
// delete temp output path
FileSystem fs = FileSystem.get(jobConf);
FileUtil.delete(fs, tempOutputPath, true);
CloudataMapReduceUtil.clearMapReduce(libDir);
}
}
public static Path putData() throws IOException {
CloudataConf nconf = new CloudataConf();
JobConf jobConf = new JobConf(ManyTableJob.class);
jobConf.set("user.name", nconf.getUserId());
String libDir = CloudataMapReduceUtil.initMapReduce(jobConf);
jobConf.setJobName("ManyTableJob_Put" + "(" + new Date() + ")");
jobConf.setLong("mapred.task.timeout", 30 * 60 * 1000);
Path outputPath = new Path("ManyTableJob_KEY_" + System.currentTimeMillis());
FileOutputFormat.setOutputPath(jobConf, outputPath);
// <MAP>
jobConf.setMapperClass(ManyTablePutMap.class);
jobConf.setInputFormat(SimpleInputFormat.class);
jobConf.setNumMapTasks(numOfTables);
jobConf.setMapSpeculativeExecution(false);
jobConf.setMaxMapAttempts(0);
// </MAP>
// <REDUCE>
jobConf.setNumReduceTasks(0);
// </REDUCE>
try {
// Run Job
JobClient.runJob(jobConf);
return outputPath;
} finally {
// delete temp output path
FileSystem fs = FileSystem.get(jobConf);
CloudataMapReduceUtil.clearMapReduce(libDir);
}
}
@Test
public void mrRun() throws Exception {
FileSystem fs = dfsCluster.getFileSystem();
Path inDir = fs.makeQualified(new Path("/user/testing/testMapperReducer/input"));
fs.delete(inDir, true);
String DATADIR = "/user/testing/testMapperReducer/data";
Path dataDir = fs.makeQualified(new Path(DATADIR));
fs.delete(dataDir, true);
Path outDir = fs.makeQualified(new Path("/user/testing/testMapperReducer/output"));
fs.delete(outDir, true);
assertTrue(fs.mkdirs(inDir));
Path INPATH = new Path(inDir, "input.txt");
OutputStream os = fs.create(INPATH);
Writer wr = new OutputStreamWriter(os, StandardCharsets.UTF_8);
wr.write(DATADIR + "/" + inputAvroFile);
wr.close();
assertTrue(fs.mkdirs(dataDir));
fs.copyFromLocalFile(new Path(DOCUMENTS_DIR, inputAvroFile), dataDir);
JobConf jobConf = getJobConf();
jobConf.set("jobclient.output.filter", "ALL");
if (ENABLE_LOCAL_JOB_RUNNER) { // enable Hadoop LocalJobRunner; this enables to run in debugger
// and set breakpoints
jobConf.set("mapred.job.tracker", "local");
}
jobConf.setMaxMapAttempts(1);
jobConf.setMaxReduceAttempts(1);
jobConf.setJar(SEARCH_ARCHIVES_JAR);
int shards = 2;
int maxReducers = Integer.MAX_VALUE;
if (ENABLE_LOCAL_JOB_RUNNER) {
// local job runner has a couple of limitations: only one reducer is supported and the
// DistributedCache doesn't work.
// see http://blog.cloudera.com/blog/2009/07/advice-on-qa-testing-your-mapreduce-jobs/
maxReducers = 1;
shards = 1;
}
String[] args =
new String[] {
"--morphline-file=" + tempDir + "/test-morphlines/solrCellDocumentTypes.conf",
"--morphline-id=morphline1",
"--solr-home-dir=" + MINIMR_CONF_DIR.getAbsolutePath(),
"--output-dir=" + outDir.toString(),
"--shards=" + shards,
"--verbose",
numRuns % 2 == 0 ? "--input-list=" + INPATH.toString() : dataDir.toString(),
numRuns % 3 == 0
? "--reducers=" + shards
: (numRuns % 3 == 1 ? "--reducers=-1" : "--reducers=" + Math.min(8, maxReducers))
};
if (numRuns % 3 == 2) {
args = concat(args, new String[] {"--fanout=2"});
}
if (numRuns == 0) {
// force (slow) MapReduce based randomization to get coverage for that as well
args =
concat(
new String[] {"-D", MapReduceIndexerTool.MAIN_MEMORY_RANDOMIZATION_THRESHOLD + "=-1"},
args);
}
MapReduceIndexerTool tool = createTool();
int res = ToolRunner.run(jobConf, tool, args);
assertEquals(0, res);
Job job = tool.job;
assertTrue(job.isComplete());
assertTrue(job.isSuccessful());
if (numRuns % 3 != 2) {
// Only run this check if mtree merge is disabled.
// With mtree merge enabled the BatchWriter counters aren't available anymore because
// variable "job" now refers to the merge job rather than the indexing job
assertEquals(
"Invalid counter "
+ SolrRecordWriter.class.getName()
+ "."
+ SolrCounters.DOCUMENTS_WRITTEN,
count,
job.getCounters()
.findCounter(SolrCounters.class.getName(), SolrCounters.DOCUMENTS_WRITTEN.toString())
.getValue());
}
// Check the output is as expected
outDir = new Path(outDir, MapReduceIndexerTool.RESULTS_DIR);
Path[] outputFiles = FileUtil.stat2Paths(fs.listStatus(outDir));
System.out.println("outputfiles:" + Arrays.toString(outputFiles));
UtilsForTests.validateSolrServerDocumentCount(MINIMR_CONF_DIR, fs, outDir, count, shards);
// run again with --dryrun mode:
tool = createTool();
args = concat(args, new String[] {"--dry-run"});
res = ToolRunner.run(jobConf, tool, args);
assertEquals(0, res);
numRuns++;
}
public JobBuilder maxMapAttempts(int maxAttempts) throws IOException {
_jobConf.setMaxMapAttempts(maxAttempts);
return this;
}