use of org.apache.hadoop.hive.metastore.api.Table in project hive by apache.
the class TestWorker method compactNoBaseLotsOfDeltas.
private void compactNoBaseLotsOfDeltas(CompactionType type) throws Exception {
conf.setIntVar(HiveConf.ConfVars.COMPACTOR_MAX_NUM_DELTA, 2);
Table t = newTable("default", "mapwb", true);
Partition p = newPartition(t, "today");
// addBaseFile(t, p, 20L, 20);
addDeltaFile(t, p, 21L, 21L, 2);
addDeltaFile(t, p, 23L, 23L, 2);
//make it look like streaming API use case
addDeltaFile(t, p, 25L, 29L, 2);
addDeltaFile(t, p, 31L, 32L, 3);
//make it looks like 31-32 has been compacted, but not cleaned
addDeltaFile(t, p, 31L, 33L, 5);
addDeltaFile(t, p, 35L, 35L, 1);
/*since COMPACTOR_MAX_NUM_DELTA=2,
we expect files 1,2 to be minor compacted by 1 job to produce delta_21_23
* 3,5 to be minor compacted by 2nd job (file 4 is obsolete) to make delta_25_33 (4th is skipped)
*
* and then the 'requested'
* minor compaction to combine delta_21_23, delta_25_33 and delta_35_35 to make delta_21_35
* or major compaction to create base_35*/
burnThroughTransactions(35);
CompactionRequest rqst = new CompactionRequest("default", "mapwb", type);
rqst.setPartitionname("ds=today");
txnHandler.compact(rqst);
startWorker();
ShowCompactResponse rsp = txnHandler.showCompact(new ShowCompactRequest());
List<ShowCompactResponseElement> compacts = rsp.getCompacts();
Assert.assertEquals(1, compacts.size());
Assert.assertEquals("ready for cleaning", compacts.get(0).getState());
FileSystem fs = FileSystem.get(conf);
FileStatus[] stat = fs.listStatus(new Path(p.getSd().getLocation()));
Assert.assertEquals(9, stat.length);
// Find the new delta file and make sure it has the right contents
BitSet matchesFound = new BitSet(9);
for (int i = 0; i < stat.length; i++) {
if (stat[i].getPath().getName().equals(makeDeltaDirName(21, 21))) {
matchesFound.set(0);
} else if (stat[i].getPath().getName().equals(makeDeltaDirName(23, 23))) {
matchesFound.set(1);
} else if (stat[i].getPath().getName().equals(makeDeltaDirNameCompacted(25, 29))) {
matchesFound.set(2);
} else if (stat[i].getPath().getName().equals(makeDeltaDirNameCompacted(31, 32))) {
matchesFound.set(3);
} else if (stat[i].getPath().getName().equals(makeDeltaDirNameCompacted(31, 33))) {
matchesFound.set(4);
} else if (stat[i].getPath().getName().equals(makeDeltaDirName(35, 35))) {
matchesFound.set(5);
} else if (stat[i].getPath().getName().equals(makeDeltaDirNameCompacted(21, 23))) {
matchesFound.set(6);
} else if (stat[i].getPath().getName().equals(makeDeltaDirNameCompacted(25, 33))) {
matchesFound.set(7);
}
switch(type) {
//yes, both do set(8)
case MINOR:
if (stat[i].getPath().getName().equals(makeDeltaDirNameCompacted(21, 35))) {
matchesFound.set(8);
}
break;
case MAJOR:
if (stat[i].getPath().getName().equals(AcidUtils.baseDir(35))) {
matchesFound.set(8);
}
break;
default:
throw new IllegalStateException();
}
}
StringBuilder sb = null;
for (int i = 0; i < stat.length; i++) {
if (!matchesFound.get(i)) {
if (sb == null) {
sb = new StringBuilder("Some files are missing at index: ");
}
sb.append(i).append(",");
}
}
if (sb != null) {
Assert.assertTrue(sb.toString(), false);
}
}
use of org.apache.hadoop.hive.metastore.api.Table in project hive by apache.
the class TestWorker method sortedPartition.
@Test
public void sortedPartition() throws Exception {
List<Order> sortCols = new ArrayList<Order>(1);
sortCols.add(new Order("b", 1));
Table t = newTable("default", "sp", true, new HashMap<String, String>(), sortCols, false);
Partition p = newPartition(t, "today", sortCols);
addBaseFile(t, p, 20L, 20);
addDeltaFile(t, p, 21L, 22L, 2);
addDeltaFile(t, p, 23L, 24L, 2);
addDeltaFile(t, p, 21L, 24L, 4);
burnThroughTransactions(25);
CompactionRequest rqst = new CompactionRequest("default", "sp", CompactionType.MINOR);
rqst.setPartitionname("ds=today");
txnHandler.compact(rqst);
startWorker();
// There should still be four directories in the location.
FileSystem fs = FileSystem.get(conf);
FileStatus[] stat = fs.listStatus(new Path(p.getSd().getLocation()));
Assert.assertEquals(4, stat.length);
}
use of org.apache.hadoop.hive.metastore.api.Table in project hive by apache.
the class TestWorker method minorPartitionWithBase.
@Test
public void minorPartitionWithBase() throws Exception {
Table t = newTable("default", "mpwb", true);
Partition p = newPartition(t, "today");
addBaseFile(t, p, 20L, 20);
addDeltaFile(t, p, 21L, 22L, 2);
addDeltaFile(t, p, 23L, 24L, 2);
burnThroughTransactions(25);
CompactionRequest rqst = new CompactionRequest("default", "mpwb", CompactionType.MINOR);
rqst.setPartitionname("ds=today");
txnHandler.compact(rqst);
startWorker();
ShowCompactResponse rsp = txnHandler.showCompact(new ShowCompactRequest());
List<ShowCompactResponseElement> compacts = rsp.getCompacts();
Assert.assertEquals(1, compacts.size());
Assert.assertEquals("ready for cleaning", compacts.get(0).getState());
// There should still be four directories in the location.
FileSystem fs = FileSystem.get(conf);
FileStatus[] stat = fs.listStatus(new Path(p.getSd().getLocation()));
Assert.assertEquals(4, stat.length);
// Find the new delta file and make sure it has the right contents
boolean sawNewDelta = false;
for (int i = 0; i < stat.length; i++) {
if (stat[i].getPath().getName().equals(makeDeltaDirNameCompacted(21, 24))) {
sawNewDelta = true;
FileStatus[] buckets = fs.listStatus(stat[i].getPath());
Assert.assertEquals(2, buckets.length);
Assert.assertTrue(buckets[0].getPath().getName().matches("bucket_0000[01]"));
Assert.assertTrue(buckets[1].getPath().getName().matches("bucket_0000[01]"));
Assert.assertEquals(208L, buckets[0].getLen());
Assert.assertEquals(208L, buckets[1].getLen());
} else {
LOG.debug("This is not the delta file you are looking for " + stat[i].getPath().getName());
}
}
Assert.assertTrue(sawNewDelta);
}
use of org.apache.hadoop.hive.metastore.api.Table in project hive by apache.
the class TestWorker method majorWithAborted.
@Test
public void majorWithAborted() throws Exception {
LOG.debug("Starting majorWithAborted");
Table t = newTable("default", "mtwb", false);
addBaseFile(t, null, 20L, 20);
addDeltaFile(t, null, 21L, 22L, 2);
addDeltaFile(t, null, 23L, 25L, 3);
addLengthFile(t, null, 23L, 25L, 3);
addDeltaFile(t, null, 26L, 27L, 2);
burnThroughTransactions(27, null, new HashSet<Long>(Arrays.asList(24L, 25L)));
CompactionRequest rqst = new CompactionRequest("default", "mtwb", CompactionType.MAJOR);
txnHandler.compact(rqst);
startWorker();
ShowCompactResponse rsp = txnHandler.showCompact(new ShowCompactRequest());
List<ShowCompactResponseElement> compacts = rsp.getCompacts();
Assert.assertEquals(1, compacts.size());
Assert.assertEquals("ready for cleaning", compacts.get(0).getState());
// There should still now be 5 directories in the location
FileSystem fs = FileSystem.get(conf);
FileStatus[] stat = fs.listStatus(new Path(t.getSd().getLocation()));
Assert.assertEquals(5, stat.length);
// Find the new delta file and make sure it has the right contents
Arrays.sort(stat);
Assert.assertEquals("base_0000027", stat[0].getPath().getName());
Assert.assertEquals("base_20", stat[1].getPath().getName());
Assert.assertEquals(makeDeltaDirName(21, 22), stat[2].getPath().getName());
Assert.assertEquals(makeDeltaDirName(23, 25), stat[3].getPath().getName());
Assert.assertEquals(makeDeltaDirName(26, 27), stat[4].getPath().getName());
}
use of org.apache.hadoop.hive.metastore.api.Table in project hive by apache.
the class TestWorker method majorTableLegacy.
@Test
public void majorTableLegacy() throws Exception {
LOG.debug("Starting majorTableLegacy");
Table t = newTable("default", "matl", false);
addLegacyFile(t, null, 20);
addDeltaFile(t, null, 21L, 22L, 2);
addDeltaFile(t, null, 23L, 24L, 2);
burnThroughTransactions(25);
CompactionRequest rqst = new CompactionRequest("default", "matl", CompactionType.MAJOR);
txnHandler.compact(rqst);
startWorker();
ShowCompactResponse rsp = txnHandler.showCompact(new ShowCompactRequest());
List<ShowCompactResponseElement> compacts = rsp.getCompacts();
Assert.assertEquals(1, compacts.size());
Assert.assertEquals("ready for cleaning", compacts.get(0).getState());
// There should still now be 5 directories in the location
FileSystem fs = FileSystem.get(conf);
FileStatus[] stat = fs.listStatus(new Path(t.getSd().getLocation()));
//Assert.assertEquals(4, stat.length);
// Find the new delta file and make sure it has the right contents
boolean sawNewBase = false;
for (int i = 0; i < stat.length; i++) {
if (stat[i].getPath().getName().equals("base_0000024")) {
sawNewBase = true;
FileStatus[] buckets = fs.listStatus(stat[i].getPath());
Assert.assertEquals(2, buckets.length);
Assert.assertTrue(buckets[0].getPath().getName().matches("bucket_0000[01]"));
Assert.assertTrue(buckets[1].getPath().getName().matches("bucket_0000[01]"));
Assert.assertEquals(624L, buckets[0].getLen());
Assert.assertEquals(624L, buckets[1].getLen());
} else {
LOG.debug("This is not the file you are looking for " + stat[i].getPath().getName());
}
}
Assert.assertTrue(sawNewBase);
}
Aggregations