Examples with ConnectionLossException org.apache.zookeeper.KeeperException.ConnectionLossException used on opensource projects

Example 1 with ConnectionLossException

use of org.apache.zookeeper.KeeperException.ConnectionLossException in project lucene-solr by apache.

the class LeaderElector method joinElection.

/**
     * Begin participating in the election process. Gets a new sequential number
     * and begins watching the node with the sequence number before it, unless it
     * is the lowest number, in which case, initiates the leader process. If the
     * node that is watched goes down, check if we are the new lowest node, else
     * watch the next lowest numbered node.
     *
     * @return sequential node number
     */
public int joinElection(ElectionContext context, boolean replacement, boolean joinAtHead) throws KeeperException, InterruptedException, IOException {
    context.joinedElectionFired();
    final String shardsElectZkPath = context.electionPath + LeaderElector.ELECTION_NODE;
    long sessionId = zkClient.getSolrZooKeeper().getSessionId();
    String id = sessionId + "-" + context.id;
    String leaderSeqPath = null;
    boolean cont = true;
    int tries = 0;
    while (cont) {
        try {
            if (joinAtHead) {
                log.debug("Node {} trying to join election at the head", id);
                List<String> nodes = OverseerTaskProcessor.getSortedElectionNodes(zkClient, shardsElectZkPath);
                if (nodes.size() < 2) {
                    leaderSeqPath = zkClient.create(shardsElectZkPath + "/" + id + "-n_", null, CreateMode.EPHEMERAL_SEQUENTIAL, false);
                } else {
                    String firstInLine = nodes.get(1);
                    log.debug("The current head: {}", firstInLine);
                    Matcher m = LEADER_SEQ.matcher(firstInLine);
                    if (!m.matches()) {
                        throw new IllegalStateException("Could not find regex match in:" + firstInLine);
                    }
                    leaderSeqPath = shardsElectZkPath + "/" + id + "-n_" + m.group(1);
                    zkClient.create(leaderSeqPath, null, CreateMode.EPHEMERAL, false);
                }
            } else {
                leaderSeqPath = zkClient.create(shardsElectZkPath + "/" + id + "-n_", null, CreateMode.EPHEMERAL_SEQUENTIAL, false);
            }
            log.debug("Joined leadership election with path: {}", leaderSeqPath);
            context.leaderSeqPath = leaderSeqPath;
            cont = false;
        } catch (ConnectionLossException e) {
            // we don't know if we made our node or not...
            List<String> entries = zkClient.getChildren(shardsElectZkPath, null, true);
            boolean foundId = false;
            for (String entry : entries) {
                String nodeId = getNodeId(entry);
                if (id.equals(nodeId)) {
                    // we did create our node...
                    foundId = true;
                    break;
                }
            }
            if (!foundId) {
                cont = true;
                if (tries++ > 20) {
                    throw new ZooKeeperException(SolrException.ErrorCode.SERVER_ERROR, "", e);
                }
                try {
                    Thread.sleep(50);
                } catch (InterruptedException e2) {
                    Thread.currentThread().interrupt();
                }
            }
        } catch (KeeperException.NoNodeException e) {
            // be working on it, lets try again
            if (tries++ > 20) {
                context = null;
                throw new ZooKeeperException(SolrException.ErrorCode.SERVER_ERROR, "", e);
            }
            cont = true;
            try {
                Thread.sleep(50);
            } catch (InterruptedException e2) {
                Thread.currentThread().interrupt();
            }
        }
    }
    checkIfIamLeader(context, replacement);
    return getSeq(context.leaderSeqPath);
}

Example 2 with ConnectionLossException

use of org.apache.zookeeper.KeeperException.ConnectionLossException in project lucene-solr by apache.

the class ZkController method getLeaderInitiatedRecoveryStateObject.

public Map<String, Object> getLeaderInitiatedRecoveryStateObject(String collection, String shardId, String coreNodeName) {
    if (collection == null || shardId == null || coreNodeName == null)
        // if we don't have complete data about a core in cloud mode, return null
        return null;
    String znodePath = getLeaderInitiatedRecoveryZnodePath(collection, shardId, coreNodeName);
    byte[] stateData = null;
    try {
        stateData = zkClient.getData(znodePath, null, new Stat(), false);
    } catch (NoNodeException ignoreMe) {
    // safe to ignore as this znode will only exist if the leader initiated recovery
    } catch (ConnectionLossException | SessionExpiredException cle) {
        // sort of safe to ignore ??? Usually these are seen when the core is going down
        // or there are bigger issues to deal with than reading this znode
        log.warn("Unable to read " + znodePath + " due to: " + cle);
    } catch (Exception exc) {
        log.error("Failed to read data from znode " + znodePath + " due to: " + exc);
        if (exc instanceof SolrException) {
            throw (SolrException) exc;
        } else {
            throw new SolrException(ErrorCode.SERVER_ERROR, "Failed to read data from znodePath: " + znodePath, exc);
        }
    }
    Map<String, Object> stateObj = null;
    if (stateData != null && stateData.length > 0) {
        // TODO: Remove later ... this is for upgrading from 4.8.x to 4.10.3 (see: SOLR-6732)
        if (stateData[0] == (byte) '{') {
            Object parsedJson = Utils.fromJSON(stateData);
            if (parsedJson instanceof Map) {
                stateObj = (Map<String, Object>) parsedJson;
            } else {
                throw new SolrException(ErrorCode.SERVER_ERROR, "Leader-initiated recovery state data is invalid! " + parsedJson);
            }
        } else {
            // old format still in ZK
            stateObj = Utils.makeMap("state", new String(stateData, StandardCharsets.UTF_8));
        }
    }
    return stateObj;
}

Example 3 with ConnectionLossException

use of org.apache.zookeeper.KeeperException.ConnectionLossException in project commons by twitter.

the class ZooKeeperClientTest method testGet.

@Test
public void testGet() throws Exception {
    final ZooKeeperClient zkClient = createZkClient();
    shutdownNetwork();
    try {
        zkClient.get(Amount.of(50L, Time.MILLISECONDS));
        fail("Expected client connection to timeout while network down");
    } catch (TimeoutException e) {
        assertTrue(zkClient.isClosed());
    }
    assertNull(zkClient.getZooKeeperClientForTests());
    final CountDownLatch blockingGetComplete = new CountDownLatch(1);
    final AtomicReference<ZooKeeper> client = new AtomicReference<ZooKeeper>();
    new Thread(new Runnable() {

        @Override
        public void run() {
            try {
                client.set(zkClient.get());
            } catch (ZooKeeperConnectionException e) {
                throw new RuntimeException(e);
            } catch (InterruptedException e) {
                throw new RuntimeException(e);
            } finally {
                blockingGetComplete.countDown();
            }
        }
    }).start();
    restartNetwork();
    // Hung blocking connects should succeed when server connection comes up
    blockingGetComplete.await();
    assertNotNull(client.get());
    // New connections should succeed now that network is back up
    long sessionId = zkClient.get().getSessionId();
    // While connected the same client should be reused (no new connections while healthy)
    assertSame(client.get(), zkClient.get());
    shutdownNetwork();
    // Our client doesn't know the network is down yet so we should be able to get()
    ZooKeeper zooKeeper = zkClient.get();
    try {
        zooKeeper.exists("/", false);
        fail("Expected client operation to fail while network down");
    } catch (ConnectionLossException e) {
    // expected
    }
    restartNetwork();
    assertEquals("Expected connection to be re-established with existing session", sessionId, zkClient.get().getSessionId());
}

Example 4 with ConnectionLossException

use of org.apache.zookeeper.KeeperException.ConnectionLossException in project helios by spotify.

the class TaskHistoryWriterTest method testZooKeeperErrorDoesntLoseItemsReally.

@Test
public void testZooKeeperErrorDoesntLoseItemsReally() throws Exception {
    final ZooKeeperClient mockClient = mock(ZooKeeperClient.class, delegatesTo(client));
    final String path = Paths.historyJobHostEventsTimestamp(JOB_ID, HOSTNAME, TIMESTAMP);
    // make save operations fail
    final AtomicBoolean throwExceptionOnCreateAndSet = new AtomicBoolean(true);
    final KeeperException exc = new ConnectionLossException();
    doAnswer(new Answer<Void>() {

        @Override
        public Void answer(InvocationOnMock invocation) throws Throwable {
            if (throwExceptionOnCreateAndSet.get()) {
                throw exc;
            } else {
                client.createAndSetData((String) invocation.getArguments()[0], (byte[]) invocation.getArguments()[1]);
                return null;
            }
        }
    }).when(mockClient).createAndSetData(path, TASK_STATUS.toJsonBytes());
    makeWriter(mockClient);
    writer.saveHistoryItem(TASK_STATUS, TIMESTAMP);
    // wait up to 10s for it to fail twice -- and make sure I mocked it correctly.
    verify(mockClient, timeout(10000).atLeast(2)).createAndSetData(path, TASK_STATUS.toJsonBytes());
    // now make the client work
    throwExceptionOnCreateAndSet.set(false);
    awaitHistoryItems();
}

Example 5 with ConnectionLossException

use of org.apache.zookeeper.KeeperException.ConnectionLossException in project zookeeper by apache.

the class InstanceManager method resetStatus.

public void resetStatus(String name) throws InterruptedException, KeeperException {
    KeeperException lastException = null;
    for (int i = 0; i < maxTries; i++) {
        try {
            zk.delete(reportsNode + '/' + name, -1);
            lastException = null;
            break;
        } catch (ConnectionLossException e) {
            lastException = e;
        } catch (NoNodeException e) {
        // great this is what we want!
        }
    }
    if (lastException != null) {
        throw lastException;
    }
}