use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.
the class TestHBaseImport method importAll.
@Test
public void importAll() throws Exception {
RawStore rdbms;
rdbms = new ObjectStore();
rdbms.setConf(conf);
String[] dbNames = new String[] { "alldb1", "alldb2" };
String[] roles = new String[] { "allrole1", "allrole2" };
String[] tokenIds = new String[] { "alltokenid1", "alltokenid2" };
String[] tokens = new String[] { "alltoken1", "alltoken2" };
String[] masterKeys = new String[] { "allmk1", "allmk2" };
int now = (int) System.currentTimeMillis() / 1000;
setupObjectStore(rdbms, roles, dbNames, tokenIds, tokens, masterKeys, now);
int baseNumRoles = store.listRoleNames() == null ? 0 : store.listRoleNames().size();
int baseNumDbs = store.getAllDatabases() == null ? 0 : store.getAllDatabases().size();
HBaseImport importer = new HBaseImport("-a");
importer.setConnections(rdbms, store);
importer.run();
for (int i = 0; i < roles.length; i++) {
Role role = store.getRole(roles[i]);
Assert.assertNotNull(role);
Assert.assertEquals(roles[i], role.getRoleName());
}
// Make sure there aren't any extra roles
Assert.assertEquals(baseNumRoles + 2, store.listRoleNames().size());
for (int i = 0; i < dbNames.length; i++) {
Database db = store.getDatabase(dbNames[i]);
Assert.assertNotNull(db);
// check one random value in the db rather than every value
Assert.assertEquals("file:/tmp", db.getLocationUri());
Table table = store.getTable(db.getName(), tableNames[0]);
Assert.assertNotNull(table);
Assert.assertEquals(now, table.getLastAccessTime());
Assert.assertEquals("input", table.getSd().getInputFormat());
table = store.getTable(db.getName(), tableNames[1]);
Assert.assertNotNull(table);
for (int j = 0; j < partVals.length; j++) {
Partition part = store.getPartition(dbNames[i], tableNames[1], Arrays.asList(partVals[j]));
Assert.assertNotNull(part);
Assert.assertEquals("file:/tmp/region=" + partVals[j], part.getSd().getLocation());
}
Assert.assertEquals(4, store.getPartitions(dbNames[i], tableNames[1], -1).size());
// Including two index table
Assert.assertEquals(4, store.getAllTables(dbNames[i]).size());
Assert.assertEquals(2, store.getIndexes(dbNames[i], tableNames[0], -1).size());
Assert.assertEquals(0, store.getIndexes(dbNames[i], tableNames[1], -1).size());
Assert.assertEquals(2, store.getFunctions(dbNames[i], "*").size());
for (int j = 0; j < funcNames.length; j++) {
Assert.assertNotNull(store.getFunction(dbNames[i], funcNames[j]));
}
}
Assert.assertEquals(baseNumDbs + 2, store.getAllDatabases().size());
// guarantee.
for (int i = 0; i < tokenIds.length; i++) {
Assert.assertEquals(tokens[i], store.getToken(tokenIds[i]));
}
String[] hbaseKeys = store.getMasterKeys();
Set<String> keys = new HashSet<>(Arrays.asList(hbaseKeys));
for (int i = 0; i < masterKeys.length; i++) {
Assert.assertTrue(keys.contains(masterKeys[i]));
}
}
use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.
the class TestHBaseImport method importOneDb.
@Test
public void importOneDb() throws Exception {
RawStore rdbms;
rdbms = new ObjectStore();
rdbms.setConf(conf);
String[] dbNames = new String[] { "onedbdb1", "onedbdb2" };
String[] roles = new String[] { "onedbrole1", "onedbrole2" };
String[] tokenIds = new String[] { "onedbtokenid1", "onedbtokenid2" };
String[] tokens = new String[] { "onedbtoken1", "onedbtoken2" };
String[] masterKeys = new String[] { "onedbmk1", "onedbmk2" };
int now = (int) System.currentTimeMillis() / 1000;
setupObjectStore(rdbms, roles, dbNames, tokenIds, tokens, masterKeys, now);
int baseNumRoles = store.listRoleNames() == null ? 0 : store.listRoleNames().size();
int baseNumDbs = store.getAllDatabases() == null ? 0 : store.getAllDatabases().size();
int baseNumToks = store.getAllTokenIdentifiers() == null ? 0 : store.getAllTokenIdentifiers().size();
int baseNumKeys = store.getMasterKeys() == null ? 0 : store.getMasterKeys().length;
HBaseImport importer = new HBaseImport("-d", dbNames[0]);
importer.setConnections(rdbms, store);
importer.run();
// Make sure there aren't any extra roles
Assert.assertEquals(baseNumRoles, store.listRoleNames().size());
Database db = store.getDatabase(dbNames[0]);
Assert.assertNotNull(db);
// check one random value in the db rather than every value
Assert.assertEquals("file:/tmp", db.getLocationUri());
Table table = store.getTable(db.getName(), tableNames[0]);
Assert.assertNotNull(table);
Assert.assertEquals(now, table.getLastAccessTime());
Assert.assertEquals("input", table.getSd().getInputFormat());
table = store.getTable(db.getName(), tableNames[1]);
Assert.assertNotNull(table);
for (int j = 0; j < partVals.length; j++) {
Partition part = store.getPartition(dbNames[0], tableNames[1], Arrays.asList(partVals[j]));
Assert.assertNotNull(part);
Assert.assertEquals("file:/tmp/region=" + partVals[j], part.getSd().getLocation());
}
Assert.assertEquals(4, store.getPartitions(dbNames[0], tableNames[1], -1).size());
// Including two index table
Assert.assertEquals(4, store.getAllTables(dbNames[0]).size());
Assert.assertEquals(2, store.getIndexes(dbNames[0], tableNames[0], -1).size());
Assert.assertEquals(0, store.getIndexes(dbNames[0], tableNames[1], -1).size());
Assert.assertEquals(2, store.getFunctions(dbNames[0], "*").size());
for (int j = 0; j < funcNames.length; j++) {
Assert.assertNotNull(store.getFunction(dbNames[0], funcNames[j]));
}
Assert.assertEquals(baseNumDbs + 1, store.getAllDatabases().size());
Assert.assertEquals(baseNumToks, store.getAllTokenIdentifiers().size());
String[] hbaseKeys = store.getMasterKeys();
Assert.assertEquals(baseNumKeys, hbaseKeys.length);
// Have to do this last as it will throw an exception
thrown.expect(NoSuchObjectException.class);
store.getDatabase(dbNames[1]);
}
use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.
the class TestHBaseImport method parallelOdd.
// Same as the test above except we create 9 of everything instead of 10. This is important
// because in using a batch size of 2 the previous test guarantees 10 /2 =5 , meaning we'll
// have 5 writes on the partition queue with exactly 2 entries. In this test we'll handle the
// case where the last entry in the queue has fewer partitions.
@Test
public void parallelOdd() throws Exception {
int parallelFactor = 9;
RawStore rdbms;
rdbms = new ObjectStore();
rdbms.setConf(conf);
String[] dbNames = new String[] { "oddparalleldb1" };
int now = (int) System.currentTimeMillis() / 1000;
for (int i = 0; i < dbNames.length; i++) {
rdbms.createDatabase(new Database(dbNames[i], "no description", "file:/tmp", emptyParameters));
List<FieldSchema> cols = new ArrayList<>();
cols.add(new FieldSchema("col1", "int", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
List<FieldSchema> partCols = new ArrayList<>();
partCols.add(new FieldSchema("region", "string", ""));
for (int j = 0; j < parallelFactor; j++) {
rdbms.createTable(new Table("t" + j, dbNames[i], "me", now, now, 0, sd, partCols, emptyParameters, null, null, null));
for (int k = 0; k < parallelFactor; k++) {
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/region=" + k);
Partition part = new Partition(Arrays.asList("p" + k), dbNames[i], "t" + j, now, now, psd, emptyParameters);
rdbms.addPartition(part);
}
}
}
HBaseImport importer = new HBaseImport("-p", "2", "-b", "2", "-d", dbNames[0]);
importer.setConnections(rdbms, store);
importer.run();
for (int i = 0; i < dbNames.length; i++) {
Database db = store.getDatabase(dbNames[i]);
Assert.assertNotNull(db);
for (int j = 0; j < parallelFactor; j++) {
Table table = store.getTable(db.getName(), "t" + j);
Assert.assertNotNull(table);
Assert.assertEquals(now, table.getLastAccessTime());
Assert.assertEquals("input", table.getSd().getInputFormat());
for (int k = 0; k < parallelFactor; k++) {
Partition part = store.getPartition(dbNames[i], "t" + j, Arrays.asList("p" + k));
Assert.assertNotNull(part);
Assert.assertEquals("file:/tmp/region=" + k, part.getSd().getLocation());
}
Assert.assertEquals(parallelFactor, store.getPartitions(dbNames[i], "t" + j, -1).size());
}
Assert.assertEquals(parallelFactor, store.getAllTables(dbNames[i]).size());
}
}
use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.
the class TestHBaseStore method createPartition.
@Test
public void createPartition() throws Exception {
String tableName = "myparttable";
int startTime = (int) (System.currentTimeMillis() / 1000);
List<FieldSchema> cols = new ArrayList<FieldSchema>();
cols.add(new FieldSchema("col1", "int", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
List<FieldSchema> partCols = new ArrayList<FieldSchema>();
partCols.add(new FieldSchema("pc", "string", ""));
Table table = new Table(tableName, DB, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
store.createTable(table);
List<String> vals = Arrays.asList("fred");
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/pc=fred");
Partition part = new Partition(vals, DB, tableName, startTime, startTime, psd, emptyParameters);
store.addPartition(part);
Partition p = store.getPartition(DB, tableName, vals);
Assert.assertEquals(1, p.getSd().getColsSize());
Assert.assertEquals("col1", p.getSd().getCols().get(0).getName());
Assert.assertEquals("int", p.getSd().getCols().get(0).getType());
Assert.assertEquals("nocomment", p.getSd().getCols().get(0).getComment());
Assert.assertEquals("serde", p.getSd().getSerdeInfo().getName());
Assert.assertEquals("seriallib", p.getSd().getSerdeInfo().getSerializationLib());
Assert.assertEquals("file:/tmp/pc=fred", p.getSd().getLocation());
Assert.assertEquals("input", p.getSd().getInputFormat());
Assert.assertEquals("output", p.getSd().getOutputFormat());
Assert.assertEquals(DB, p.getDbName());
Assert.assertEquals(tableName, p.getTableName());
Assert.assertEquals(1, p.getValuesSize());
Assert.assertEquals("fred", p.getValues().get(0));
Assert.assertTrue(store.doesPartitionExist(DB, tableName, vals));
Assert.assertFalse(store.doesPartitionExist(DB, tableName, Arrays.asList("bob")));
}
use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.
the class TestHBaseStoreBitVector method createMockTableAndPartition.
private Table createMockTableAndPartition(String partType, String partVal) throws Exception {
List<FieldSchema> cols = new ArrayList<FieldSchema>();
cols.add(new FieldSchema("col1", partType, ""));
List<String> vals = new ArrayList<String>();
vals.add(partVal);
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
Map<String, String> params = new HashMap<String, String>();
params.put("key", "value");
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 17, serde, Arrays.asList("bucketcol"), Arrays.asList(new Order("sortcol", 1)), params);
int currentTime = (int) (System.currentTimeMillis() / 1000);
Table table = new Table(TBL, DB, "me", currentTime, currentTime, 0, sd, cols, emptyParameters, null, null, null);
store.createTable(table);
Partition part = new Partition(vals, DB, TBL, currentTime, currentTime, sd, emptyParameters);
store.addPartition(part);
return table;
}
Aggregations