private void runIncrementalPELoad(Configuration conf, HTable table, Path outDir)
throws Exception {
Job job = new Job(conf, "testLocalMRIncrementalLoad");
job.setWorkingDirectory(util.getDataTestDirOnTestFS("runIncrementalPELoad"));
job.getConfiguration()
.setStrings(
"io.serializations",
conf.get("io.serializations"),
MutationSerialization.class.getName(),
ResultSerialization.class.getName(),
KeyValueSerialization.class.getName());
setupRandomGeneratorMapper(job);
HFileOutputFormat2.configureIncrementalLoad(
job, table.getTableDescriptor(), table.getRegionLocator());
FileOutputFormat.setOutputPath(job, outDir);
Assert.assertFalse(util.getTestFileSystem().exists(outDir));
assertEquals(table.getRegionLocator().getAllRegionLocations().size(), job.getNumReduceTasks());
assertTrue(job.waitForCompletion(true));
}
private void doIncrementalLoadTest(boolean shouldChangeRegions) throws Exception {
util = new HBaseTestingUtility();
Configuration conf = util.getConfiguration();
byte[][] splitKeys = generateRandomSplitKeys(4);
util.startMiniCluster();
try {
HTable table = util.createTable(TABLE_NAME, FAMILIES, splitKeys);
Admin admin = table.getConnection().getAdmin();
Path testDir = util.getDataTestDirOnTestFS("testLocalMRIncrementalLoad");
assertEquals("Should start with empty table", 0, util.countRows(table));
int numRegions = -1;
try (RegionLocator r = table.getRegionLocator()) {
numRegions = r.getStartKeys().length;
}
assertEquals("Should make 5 regions", numRegions, 5);
// Generate the bulk load files
util.startMiniMapReduceCluster();
runIncrementalPELoad(conf, table.getTableDescriptor(), table.getRegionLocator(), testDir);
// This doesn't write into the table, just makes files
assertEquals("HFOF should not touch actual table", 0, util.countRows(table));
// Make sure that a directory was created for every CF
int dir = 0;
for (FileStatus f : testDir.getFileSystem(conf).listStatus(testDir)) {
for (byte[] family : FAMILIES) {
if (Bytes.toString(family).equals(f.getPath().getName())) {
++dir;
}
}
}
assertEquals("Column family not found in FS.", FAMILIES.length, dir);
// handle the split case
if (shouldChangeRegions) {
LOG.info("Changing regions in table");
admin.disableTable(table.getName());
while (util.getMiniHBaseCluster()
.getMaster()
.getAssignmentManager()
.getRegionStates()
.isRegionsInTransition()) {
Threads.sleep(200);
LOG.info("Waiting on table to finish disabling");
}
util.deleteTable(table.getName());
byte[][] newSplitKeys = generateRandomSplitKeys(14);
table = util.createTable(TABLE_NAME, FAMILIES, newSplitKeys);
while (table.getRegionLocator().getAllRegionLocations().size() != 15
|| !admin.isTableAvailable(table.getName())) {
Thread.sleep(200);
LOG.info("Waiting for new region assignment to happen");
}
}
// Perform the actual load
new LoadIncrementalHFiles(conf).doBulkLoad(testDir, table);
// Ensure data shows up
int expectedRows = NMapInputFormat.getNumMapTasks(conf) * ROWSPERSPLIT;
assertEquals(
"LoadIncrementalHFiles should put expected data in table",
expectedRows,
util.countRows(table));
Scan scan = new Scan();
ResultScanner results = table.getScanner(scan);
for (Result res : results) {
assertEquals(FAMILIES.length, res.rawCells().length);
Cell first = res.rawCells()[0];
for (Cell kv : res.rawCells()) {
assertTrue(CellUtil.matchingRow(first, kv));
assertTrue(Bytes.equals(CellUtil.cloneValue(first), CellUtil.cloneValue(kv)));
}
}
results.close();
String tableDigestBefore = util.checksumRows(table);
// Cause regions to reopen
admin.disableTable(TABLE_NAME);
while (!admin.isTableDisabled(TABLE_NAME)) {
Thread.sleep(200);
LOG.info("Waiting for table to disable");
}
admin.enableTable(TABLE_NAME);
util.waitTableAvailable(TABLE_NAME);
assertEquals(
"Data should remain after reopening of regions",
tableDigestBefore,
util.checksumRows(table));
} finally {
util.shutdownMiniMapReduceCluster();
util.shutdownMiniCluster();
}
}
