@Override
void perform() throws IOException {
Admin admin = connection.getAdmin();
try {
HTableDescriptor htd = createTableDesc();
TableName tableName = htd.getTableName();
if (admin.tableExists(tableName)) {
return;
}
String numRegionKey = String.format(NUM_REGIONS_KEY, this.getClass().getSimpleName());
numRegions = getConf().getInt(numRegionKey, DEFAULT_NUM_REGIONS);
byte[] startKey = Bytes.toBytes("row-0000000000");
byte[] endKey = Bytes.toBytes("row-" + Integer.MAX_VALUE);
LOG.info("Creating table:" + htd);
admin.createTable(htd, startKey, endKey, numRegions);
Assert.assertTrue("Table: " + htd + " was not created", admin.tableExists(tableName));
HTableDescriptor freshTableDesc = admin.getTableDescriptor(tableName);
enabledTables.put(tableName, freshTableDesc);
LOG.info("Created table:" + freshTableDesc);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
throw e;
} finally {
admin.close();
}
verifyTables();
}
public static void CreateTable(HBaseConfiguration conf) throws IOException {
HBaseAdmin admin = new HBaseAdmin(conf);
HTableDescriptor tableDescriptor = new HTableDescriptor(TableName.valueOf("people"));
tableDescriptor.addFamily(new HColumnDescriptor("personal"));
tableDescriptor.addFamily(new HColumnDescriptor("contactinfo"));
tableDescriptor.addFamily(new HColumnDescriptor("creditcard"));
admin.createTable(tableDescriptor);
}
private HTableDescriptor createTableDesc() {
String tableName =
"ittable-" + String.format("%010d", RandomUtils.nextInt(Integer.MAX_VALUE));
String familyName = "cf-" + Math.abs(RandomUtils.nextInt());
HTableDescriptor htd = new HTableDescriptor(TableName.valueOf(tableName));
// add random column family
htd.addFamily(new HColumnDescriptor(familyName));
return htd;
}
@Override
public HTableDescriptor createHtd(final String tableName) {
HTableDescriptor htd = new HTableDescriptor(tableName);
HColumnDescriptor hcd = new HColumnDescriptor(TEST_FAMILY);
hcd.setMobEnabled(true);
hcd.setMobThreshold(0L);
htd.addFamily(hcd);
return htd;
}
public void createTable(String tableName, List<String> columnFamilies) {
try {
Admin admin = connection.getAdmin();
HTableDescriptor descriptor = new HTableDescriptor(TableName.valueOf(tableName));
for (String family : columnFamilies) {
descriptor.addFamily(new HColumnDescriptor(family));
}
admin.createTable(descriptor);
} catch (IOException e) {
e.printStackTrace();
}
}
@Override
void perform() throws IOException {
HTableDescriptor selected = selectTable(disabledTables);
if (selected == null) {
return;
}
HColumnDescriptor columnDesc = selectFamily(selected);
if (columnDesc == null) {
return;
}
Admin admin = connection.getAdmin();
int versions = RandomUtils.nextInt(10) + 3;
try {
TableName tableName = selected.getTableName();
LOG.info(
"Altering versions of column family: "
+ columnDesc
+ " to: "
+ versions
+ " in table: "
+ tableName);
columnDesc.setMinVersions(versions);
columnDesc.setMaxVersions(versions);
admin.modifyTable(tableName, selected);
// assertion
HTableDescriptor freshTableDesc = admin.getTableDescriptor(tableName);
HColumnDescriptor freshColumnDesc = freshTableDesc.getFamily(columnDesc.getName());
Assert.assertEquals(
"Column family: " + columnDesc + " was not altered",
freshColumnDesc.getMaxVersions(),
versions);
Assert.assertEquals(
"Column family: " + freshColumnDesc + " was not altered",
freshColumnDesc.getMinVersions(),
versions);
LOG.info(
"Altered versions of column family: "
+ columnDesc
+ " to: "
+ versions
+ " in table: "
+ tableName);
disabledTables.put(tableName, freshTableDesc);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
throw e;
} finally {
admin.close();
}
verifyTables();
}
@Override
void perform() throws IOException {
HTableDescriptor selected = selectTable(disabledTables);
if (selected == null) {
return;
}
HColumnDescriptor columnDesc = selectFamily(selected);
if (columnDesc == null) {
return;
}
Admin admin = connection.getAdmin();
try {
TableName tableName = selected.getTableName();
// possible DataBlockEncoding ids
int[] possibleIds = {0, 2, 3, 4, 6};
short id = (short) possibleIds[RandomUtils.nextInt(possibleIds.length)];
LOG.info(
"Altering encoding of column family: "
+ columnDesc
+ " to: "
+ id
+ " in table: "
+ tableName);
columnDesc.setDataBlockEncoding(DataBlockEncoding.getEncodingById(id));
admin.modifyTable(tableName, selected);
// assertion
HTableDescriptor freshTableDesc = admin.getTableDescriptor(tableName);
HColumnDescriptor freshColumnDesc = freshTableDesc.getFamily(columnDesc.getName());
Assert.assertEquals(
"Encoding of column family: " + columnDesc + " was not altered",
freshColumnDesc.getDataBlockEncoding().getId(),
id);
LOG.info(
"Altered encoding of column family: "
+ freshColumnDesc
+ " to: "
+ id
+ " in table: "
+ tableName);
disabledTables.put(tableName, freshTableDesc);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
throw e;
} finally {
admin.close();
}
verifyTables();
}
// populate tables
@Override
void perform() throws IOException {
HTableDescriptor selected = selectTable(enabledTables);
if (selected == null) {
return;
}
Admin admin = connection.getAdmin();
TableName tableName = selected.getTableName();
try (Table table = connection.getTable(tableName)) {
ArrayList<HRegionInfo> regionInfos =
new ArrayList<HRegionInfo>(admin.getTableRegions(selected.getTableName()));
int numRegions = regionInfos.size();
// average number of rows to be added per action to each region
int average_rows = 1;
int numRows = average_rows * numRegions;
LOG.info("Adding " + numRows + " rows to table: " + selected);
for (int i = 0; i < numRows; i++) {
// nextInt(Integer.MAX_VALUE)) to return positive numbers only
byte[] rowKey =
Bytes.toBytes(
"row-" + String.format("%010d", RandomUtils.nextInt(Integer.MAX_VALUE)));
HColumnDescriptor cfd = selectFamily(selected);
if (cfd == null) {
return;
}
byte[] family = cfd.getName();
byte[] qualifier = Bytes.toBytes("col-" + RandomUtils.nextInt(Integer.MAX_VALUE) % 10);
byte[] value = Bytes.toBytes("val-" + RandomStringUtils.randomAlphanumeric(10));
Put put = new Put(rowKey);
put.addColumn(family, qualifier, value);
table.put(put);
}
HTableDescriptor freshTableDesc = admin.getTableDescriptor(tableName);
enabledTables.put(tableName, freshTableDesc);
LOG.info("Added " + numRows + " rows to table: " + selected);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
throw e;
} finally {
admin.close();
}
verifyTables();
}
@Override
void perform() throws IOException {
HTableDescriptor selected = selectTable(disabledTables);
if (selected == null) {
return;
}
Admin admin = connection.getAdmin();
try {
TableName tableName = selected.getTableName();
LOG.info("Enabling table :" + selected);
admin.enableTable(tableName);
Assert.assertTrue(
"Table: " + selected + " was not enabled", admin.isTableEnabled(tableName));
HTableDescriptor freshTableDesc = admin.getTableDescriptor(tableName);
enabledTables.put(tableName, freshTableDesc);
LOG.info("Enabled table :" + freshTableDesc);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
// TODO workaround
// loose restriction for TableNotDisabledException/TableNotEnabledException thrown in sync
// operations 1) when enable/disable starts, the table state is changed to
// ENABLING/DISABLING (ZK node in 1.x), which will be further changed to ENABLED/DISABLED
// once the operation completes 2) if master failover happens in the middle of the
// enable/disable operation, the new master will try to recover the tables in
// ENABLING/DISABLING state, as programmed in
// AssignmentManager#recoverTableInEnablingState() and
// AssignmentManager#recoverTableInDisablingState()
// 3) after the new master initialization completes, the procedure tries to re-do the
// enable/disable operation, which was already done. Ignore those exceptions before
// change of behaviors of AssignmentManager in presence of PV2
if (e instanceof TableNotDisabledException) {
LOG.warn("Caught TableNotDisabledException in action: " + this.getClass());
e.printStackTrace();
} else {
throw e;
}
} finally {
admin.close();
}
verifyTables();
}
// ColumnAction has implemented selectFamily() shared by multiple family Actions
protected HColumnDescriptor selectFamily(HTableDescriptor htd) {
if (htd == null) {
return null;
}
HColumnDescriptor[] families = htd.getColumnFamilies();
if (families.length == 0) {
LOG.info("No column families in table: " + htd);
return null;
}
HColumnDescriptor randomCfd = families[RandomUtils.nextInt(families.length)];
return randomCfd;
}
/** Create the Mob Table. */
public static void createMobTable(
final HBaseTestingUtility util,
final TableName tableName,
int regionReplication,
final byte[]... families)
throws IOException, InterruptedException {
HTableDescriptor htd = new HTableDescriptor(tableName);
htd.setRegionReplication(regionReplication);
for (byte[] family : families) {
HColumnDescriptor hcd = new HColumnDescriptor(family);
hcd.setMobEnabled(true);
hcd.setMobThreshold(0L);
htd.addFamily(hcd);
}
byte[][] splitKeys = SnapshotTestingUtils.getSplitKeys();
util.getHBaseAdmin().createTable(htd, splitKeys);
SnapshotTestingUtils.waitForTableToBeOnline(util, tableName);
assertEquals(
(splitKeys.length + 1) * regionReplication,
util.getHBaseAdmin().getTableRegions(tableName).size());
}
@Override
void perform() throws IOException {
HTableDescriptor selected = selectTable(disabledTables);
HColumnDescriptor cfd = selectFamily(selected);
if (selected == null || cfd == null) {
return;
}
Admin admin = connection.getAdmin();
try {
if (selected.getColumnFamilies().length < 2) {
LOG.info("No enough column families to delete in table " + selected.getTableName());
return;
}
TableName tableName = selected.getTableName();
LOG.info("Deleting column family: " + cfd + " from table: " + tableName);
admin.deleteColumnFamily(tableName, cfd.getName());
// assertion
HTableDescriptor freshTableDesc = admin.getTableDescriptor(tableName);
Assert.assertFalse(
"Column family: " + cfd + " was not added", freshTableDesc.hasFamily(cfd.getName()));
LOG.info("Deleted column family: " + cfd + " from table: " + tableName);
disabledTables.put(tableName, freshTableDesc);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
throw e;
} finally {
admin.close();
}
verifyTables();
}
@Override
void perform() throws IOException {
HTableDescriptor selected = selectTable(disabledTables);
if (selected == null) {
return;
}
Admin admin = connection.getAdmin();
try {
HColumnDescriptor cfd = createFamilyDesc();
if (selected.hasFamily(cfd.getName())) {
LOG.info(
new String(cfd.getName()) + " already exists in table " + selected.getTableName());
return;
}
TableName tableName = selected.getTableName();
LOG.info("Adding column family: " + cfd + " to table: " + tableName);
admin.addColumn(tableName, cfd);
// assertion
HTableDescriptor freshTableDesc = admin.getTableDescriptor(tableName);
Assert.assertTrue(
"Column family: " + cfd + " was not added", freshTableDesc.hasFamily(cfd.getName()));
LOG.info("Added column family: " + cfd + " to table: " + tableName);
disabledTables.put(tableName, freshTableDesc);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
throw e;
} finally {
admin.close();
}
verifyTables();
}
/**
* Create a Mob table.
*
* @param util
* @param tableName
* @param families
* @return An HTable instance for the created table.
* @throws IOException
*/
public static Table createMobTable(
final HBaseTestingUtility util, final TableName tableName, final byte[]... families)
throws IOException {
HTableDescriptor htd = new HTableDescriptor(tableName);
for (byte[] family : families) {
HColumnDescriptor hcd = new HColumnDescriptor(family);
// Disable blooms (they are on by default as of 0.95) but we disable them
// here because
// tests have hard coded counts of what to expect in block cache, etc.,
// and blooms being
// on is interfering.
hcd.setBloomFilterType(BloomType.NONE);
hcd.setMobEnabled(true);
hcd.setMobThreshold(0L);
htd.addFamily(hcd);
}
util.getHBaseAdmin().createTable(htd);
// HBaseAdmin only waits for regions to appear in hbase:meta we should wait
// until they are assigned
util.waitUntilAllRegionsAssigned(htd.getTableName());
return ConnectionFactory.createConnection(util.getConfiguration()).getTable(htd.getTableName());
}
@Override
void perform() throws IOException {
HTableDescriptor selected = selectTable(disabledTables);
if (selected == null) {
return;
}
Admin admin = connection.getAdmin();
try {
TableName tableName = selected.getTableName();
LOG.info("Deleting table :" + selected);
admin.deleteTable(tableName);
Assert.assertFalse("Table: " + selected + " was not deleted", admin.tableExists(tableName));
deletedTables.put(tableName, selected);
LOG.info("Deleted table :" + selected);
} catch (Exception e) {
LOG.warn("Caught exception in action: " + this.getClass());
throw e;
} finally {
admin.close();
}
verifyTables();
}
@Test
public void testHostRank() throws Exception {
if (System.getProperty("prop.mapred.job.tracker") != null) {
if (LOG.isInfoEnabled())
LOG.info("testHBaseInputOutput: Ignore this test if not local mode.");
return;
}
File jarTest = new File(System.getProperty("prop.jarLocation"));
if (!jarTest.exists()) {
fail(
"Could not find Giraph jar at "
+ "location specified by 'prop.jarLocation'. "
+ "Make sure you built the main Giraph artifact?.");
}
MiniHBaseCluster cluster = null;
MiniZooKeeperCluster zkCluster = null;
FileSystem fs = null;
try {
// using the restart method allows us to avoid having the hbase
// root directory overwritten by /home/$username
zkCluster = testUtil.startMiniZKCluster();
testUtil.restartHBaseCluster(2);
cluster = testUtil.getMiniHBaseCluster();
final byte[] OL_BYTES = Bytes.toBytes("ol");
final byte[] S_BYTES = Bytes.toBytes("s");
final byte[] METADATA_BYTES = Bytes.toBytes("mtdt");
final byte[] HR_BYTES = Bytes.toBytes("_hr_");
final byte[] TAB = Bytes.toBytes(TABLE_NAME);
Configuration conf = cluster.getConfiguration();
HTableDescriptor desc = new HTableDescriptor(TAB);
desc.addFamily(new HColumnDescriptor(OL_BYTES));
desc.addFamily(new HColumnDescriptor(S_BYTES));
desc.addFamily(new HColumnDescriptor(METADATA_BYTES));
HBaseAdmin hbaseAdmin = new HBaseAdmin(conf);
if (hbaseAdmin.isTableAvailable(TABLE_NAME)) {
hbaseAdmin.disableTable(TABLE_NAME);
hbaseAdmin.deleteTable(TABLE_NAME);
}
hbaseAdmin.createTable(desc);
/**
* Enter the initial data (a,b), (b,c), (a,c) a = 1.0 - google b = 1.0 - yahoo c = 1.0 - bing
*/
HTable table = new HTable(conf, TABLE_NAME);
Put p1 = new Put(Bytes.toBytes("com.google.www"));
p1.add(OL_BYTES, Bytes.toBytes("www.yahoo.com"), Bytes.toBytes("ab"));
Put p2 = new Put(Bytes.toBytes("com.google.www"));
p2.add(OL_BYTES, Bytes.toBytes("www.bing.com"), Bytes.toBytes("ac"));
p2.add(OL_BYTES, Bytes.toBytes("www.bing.com"), Bytes.toBytes("invalid1"));
p2.add(OL_BYTES, Bytes.toBytes("www.google.com"), Bytes.toBytes("invalid2"));
Put p3 = new Put(Bytes.toBytes("com.yahoo.www"));
p3.add(OL_BYTES, Bytes.toBytes("www.bing.com"), Bytes.toBytes("bc"));
// p3.add(OL_BYTES, Bytes.toBytes(""), Bytes.toBytes("invalid4"));
Put p4 = new Put(Bytes.toBytes("com.bing.www"));
// TODO: Handle below case. use apache isValid method.
p4.add(OL_BYTES, Bytes.toBytes("http://invalidurl"), Bytes.toBytes("invalid5"));
p4.add(S_BYTES, S_BYTES, Bytes.toBytes(10.0d));
Put p5 = new Put(Bytes.toBytes("dummy"));
p5.add(S_BYTES, S_BYTES, Bytes.toBytes(10.0d));
table.put(p1);
table.put(p2);
table.put(p3);
table.put(p4);
table.put(p5);
// Set Giraph configuration
// now operate over HBase using Vertex I/O formats
conf.set(TableInputFormat.INPUT_TABLE, TABLE_NAME);
conf.set(TableOutputFormat.OUTPUT_TABLE, TABLE_NAME);
// Start the giraph job
GiraphJob giraphJob = new GiraphJob(conf, BspCase.getCallingMethodName());
GiraphConfiguration giraphConf = giraphJob.getConfiguration();
giraphConf.setZooKeeperConfiguration(cluster.getMaster().getZooKeeper().getQuorum());
setupConfiguration(giraphJob);
giraphConf.setComputationClass(LinkRankComputation.class);
giraphConf.setMasterComputeClass(LinkRankVertexMasterCompute.class);
giraphConf.setOutEdgesClass(ByteArrayEdges.class);
giraphConf.setVertexInputFormatClass(Nutch2HostInputFormat.class);
giraphConf.setVertexOutputFormatClass(Nutch2HostOutputFormat.class);
giraphConf.setInt("giraph.linkRank.superstepCount", 10);
giraphConf.setInt("giraph.linkRank.scale", 10);
giraphConf.set("giraph.linkRank.family", "mtdt");
giraphConf.set("giraph.linkRank.qualifier", "_hr_");
giraphConf.setVertexInputFilterClass(HostRankVertexFilter.class);
assertTrue(giraphJob.run(true));
if (LOG.isInfoEnabled()) LOG.info("Giraph job successful. Checking output qualifier.");
/** Check the results * */
Result result;
String key;
byte[] calculatedScoreByte;
HashMap expectedValues = new HashMap<String, Double>();
expectedValues.put("com.google.www", 1.3515060339386287d);
expectedValues.put("com.yahoo.www", 4.144902009567587d);
expectedValues.put("com.bing.www", 9.063893290511482d);
for (Object keyObject : expectedValues.keySet()) {
key = keyObject.toString();
result = table.get(new Get(key.getBytes()));
calculatedScoreByte = result.getValue(METADATA_BYTES, HR_BYTES);
assertNotNull(calculatedScoreByte);
assertTrue(calculatedScoreByte.length > 0);
Assert.assertEquals(
"Scores are not the same",
(Double) expectedValues.get(key),
Bytes.toDouble(calculatedScoreByte),
DELTA);
}
} finally {
if (cluster != null) {
cluster.shutdown();
}
if (zkCluster != null) {
zkCluster.shutdown();
}
// clean test files
if (fs != null) {
fs.delete(hbaseRootdir);
}
}
}
