Examples with HRegionServer org.apache.hadoop.hbase.regionserver.HRegionServer used on opensource projects

Example 51 with HRegionServer

use of org.apache.hadoop.hbase.regionserver.HRegionServer in project hbase by apache.

the class TestSerialReplication method testRegionMove.

@Test
public void testRegionMove() throws Exception {
    TableName tableName = createTable();
    try (Table table = UTIL.getConnection().getTable(tableName)) {
        for (int i = 0; i < 100; i++) {
            table.put(new Put(Bytes.toBytes(i)).addColumn(CF, CQ, Bytes.toBytes(i)));
        }
    }
    RegionInfo region = UTIL.getAdmin().getRegions(tableName).get(0);
    HRegionServer rs = UTIL.getOtherRegionServer(UTIL.getRSForFirstRegionInTable(tableName));
    moveRegion(region, rs);
    try (Table table = UTIL.getConnection().getTable(tableName)) {
        for (int i = 100; i < 200; i++) {
            table.put(new Put(Bytes.toBytes(i)).addColumn(CF, CQ, Bytes.toBytes(i)));
        }
    }
    enablePeerAndWaitUntilReplicationDone(200);
    checkOrder(200);
}

Example 52 with HRegionServer

use of org.apache.hadoop.hbase.regionserver.HRegionServer in project hbase by apache.

the class TestSyncReplicationMoreLogsInLocalCopyToRemote method testSplitLog.

@Test
public void testSplitLog() throws Exception {
    UTIL1.getAdmin().disableReplicationPeer(PEER_ID);
    UTIL2.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.STANDBY);
    UTIL1.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.ACTIVE);
    HRegionServer rs = UTIL1.getRSForFirstRegionInTable(TABLE_NAME);
    DualAsyncFSWALForTest wal = (DualAsyncFSWALForTest) rs.getWAL(RegionInfoBuilder.newBuilder(TABLE_NAME).build());
    wal.setRemoteBroken();
    try (AsyncConnection conn = ConnectionFactory.createAsyncConnection(UTIL1.getConfiguration()).get()) {
        AsyncTable<?> table = conn.getTableBuilder(TABLE_NAME).setMaxAttempts(1).build();
        try {
            table.put(new Put(Bytes.toBytes(0)).addColumn(CF, CQ, Bytes.toBytes(0))).get();
            fail("Should fail since the rs will crash and we will not retry");
        } catch (ExecutionException e) {
            // expected
            LOG.info("Expected error:", e);
        }
    }
    UTIL1.waitFor(60000, new ExplainingPredicate<Exception>() {

        @Override
        public boolean evaluate() throws Exception {
            try (Table table = UTIL1.getConnection().getTable(TABLE_NAME)) {
                return table.exists(new Get(Bytes.toBytes(0)));
            }
        }

        @Override
        public String explainFailure() throws Exception {
            return "The row is still not available";
        }
    });
    UTIL2.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.DOWNGRADE_ACTIVE);
    // We should have copied the local log to remote, so we should be able to get the value
    try (Table table = UTIL2.getConnection().getTable(TABLE_NAME)) {
        assertEquals(0, Bytes.toInt(table.get(new Get(Bytes.toBytes(0))).getValue(CF, CQ)));
    }
}

Example 53 with HRegionServer

use of org.apache.hadoop.hbase.regionserver.HRegionServer in project hbase by apache.

the class TestDrainReplicationQueuesForStandBy method test.

@Test
public void test() throws Exception {
    UTIL2.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.STANDBY);
    UTIL1.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.ACTIVE);
    UTIL1.getAdmin().disableReplicationPeer(PEER_ID);
    write(UTIL1, 0, 100);
    HRegionServer rs = UTIL1.getRSForFirstRegionInTable(TABLE_NAME);
    String walGroupId = AbstractFSWALProvider.getWALPrefixFromWALName(((AbstractFSWAL<?>) rs.getWAL(RegionInfoBuilder.newBuilder(TABLE_NAME).build())).getCurrentFileName().getName());
    UTIL2.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.DOWNGRADE_ACTIVE);
    // transit cluster2 to DA and cluster 1 to S
    verify(UTIL2, 0, 100);
    UTIL1.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.STANDBY);
    // delete the original value, and then major compact
    try (Table table = UTIL2.getConnection().getTable(TABLE_NAME)) {
        for (int i = 0; i < 100; i++) {
            table.delete(new Delete(Bytes.toBytes(i)));
        }
    }
    UTIL2.flush(TABLE_NAME);
    UTIL2.compact(TABLE_NAME, true);
    // wait until the new values are replicated back to cluster1
    HRegion region = rs.getRegions(TABLE_NAME).get(0);
    UTIL1.waitFor(30000, new ExplainingPredicate<Exception>() {

        @Override
        public boolean evaluate() throws Exception {
            return region.get(new Get(Bytes.toBytes(99))).isEmpty();
        }

        @Override
        public String explainFailure() throws Exception {
            return "Replication has not been catched up yet";
        }
    });
    // transit cluster1 to DA and cluster2 to S, then we will start replicating from cluster1 to
    // cluster2
    UTIL1.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.DOWNGRADE_ACTIVE);
    UTIL2.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID, SyncReplicationState.STANDBY);
    UTIL1.getAdmin().enableReplicationPeer(PEER_ID);
    // confirm that we will not replicate the old data which causes inconsistency
    ReplicationSource source = (ReplicationSource) ((Replication) rs.getReplicationSourceService()).getReplicationManager().getSource(PEER_ID);
    UTIL1.waitFor(30000, new ExplainingPredicate<Exception>() {

        @Override
        public boolean evaluate() throws Exception {
            return !source.workerThreads.containsKey(walGroupId);
        }

        @Override
        public String explainFailure() throws Exception {
            return "Replication has not been catched up yet";
        }
    });
    HRegion region2 = UTIL2.getMiniHBaseCluster().getRegions(TABLE_NAME).get(0);
    for (int i = 0; i < 100; i++) {
        assertTrue(region2.get(new Get(Bytes.toBytes(i))).isEmpty());
    }
}

Example 54 with HRegionServer

use of org.apache.hadoop.hbase.regionserver.HRegionServer in project hbase by apache.

the class TestRefreshRecoveredReplication method testReplicationRefreshSource.

@Test
public void testReplicationRefreshSource() throws Exception {
    // put some data
    for (int i = 0; i < BATCH; i++) {
        byte[] r = Bytes.toBytes(i);
        table1.put(new Put(r).addColumn(famName, famName, r));
    }
    // Kill rs holding table region. There are only TWO servers. We depend on it.
    List<RegionServerThread> rss = UTIL1.getMiniHBaseCluster().getLiveRegionServerThreads();
    assertEquals(2, rss.size());
    Optional<RegionServerThread> server = rss.stream().filter(rst -> CollectionUtils.isNotEmpty(rst.getRegionServer().getRegions(tablename))).findAny();
    Assert.assertTrue(server.isPresent());
    HRegionServer otherServer = rss.get(0).getRegionServer() == server.get().getRegionServer() ? rss.get(1).getRegionServer() : rss.get(0).getRegionServer();
    server.get().getRegionServer().abort("stopping for test");
    // waiting for recovered peer to appear.
    Replication replication = (Replication) otherServer.getReplicationSourceService();
    UTIL1.waitFor(60000, () -> !replication.getReplicationManager().getOldSources().isEmpty());
    // Wait on only one server being up.
    UTIL1.waitFor(60000, () -> UTIL1.getMiniHBaseCluster().getLiveRegionServerThreads().size() == NUM_SLAVES1 - 1);
    UTIL1.waitTableAvailable(tablename);
    LOG.info("Available {}", tablename);
    // disable peer to trigger refreshSources
    hbaseAdmin.disableReplicationPeer(PEER_ID2);
    LOG.info("has replicated {} rows before disable peer", checkReplicationData());
    hbaseAdmin.enableReplicationPeer(PEER_ID2);
    // waiting to replicate all data to slave
    UTIL2.waitFor(60000, () -> {
        int count = checkReplicationData();
        LOG.info("Waiting all logs pushed to slave. Expected {} , actual {}", BATCH, count);
        return count == BATCH;
    });
}

Example 55 with HRegionServer

use of org.apache.hadoop.hbase.regionserver.HRegionServer in project hbase by apache.

the class HBaseTestingUtil method getRSForFirstRegionInTable.

/**
 * Tool to get the reference to the region server object that holds the region of the specified
 * user table.
 * @param tableName user table to lookup in hbase:meta
 * @return region server that holds it, null if the row doesn't exist
 */
public HRegionServer getRSForFirstRegionInTable(TableName tableName) throws IOException, InterruptedException {
    List<RegionInfo> regions = getAdmin().getRegions(tableName);
    if (regions == null || regions.isEmpty()) {
        return null;
    }
    LOG.debug("Found " + regions.size() + " regions for table " + tableName);
    byte[] firstRegionName = regions.stream().filter(r -> !r.isOffline()).map(RegionInfo::getRegionName).findFirst().orElseThrow(() -> new IOException("online regions not found in table " + tableName));
    LOG.debug("firstRegionName=" + Bytes.toString(firstRegionName));
    long pause = getConfiguration().getLong(HConstants.HBASE_CLIENT_PAUSE, HConstants.DEFAULT_HBASE_CLIENT_PAUSE);
    int numRetries = getConfiguration().getInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER, HConstants.DEFAULT_HBASE_CLIENT_RETRIES_NUMBER);
    RetryCounter retrier = new RetryCounter(numRetries + 1, (int) pause, TimeUnit.MICROSECONDS);
    while (retrier.shouldRetry()) {
        int index = getMiniHBaseCluster().getServerWith(firstRegionName);
        if (index != -1) {
            return getMiniHBaseCluster().getRegionServerThreads().get(index).getRegionServer();
        }
        // Came back -1. Region may not be online yet. Sleep a while.
        retrier.sleepUntilNextRetry();
    }
    return null;
}