Examples with NoShardAvailableActionException org.elasticsearch.action.NoShardAvailableActionException used on opensource projects

Example 1 with NoShardAvailableActionException

use of org.elasticsearch.action.NoShardAvailableActionException in project elasticsearch by elastic.

the class BroadcastReplicationTests method testNotStartedPrimary.

public void testNotStartedPrimary() throws InterruptedException, ExecutionException, IOException {
    final String index = "test";
    setState(clusterService, state(index, randomBoolean(), randomBoolean() ? ShardRoutingState.INITIALIZING : ShardRoutingState.UNASSIGNED, ShardRoutingState.UNASSIGNED));
    logger.debug("--> using initial state:\n{}", clusterService.state());
    Future<BroadcastResponse> response = (broadcastReplicationAction.execute(new DummyBroadcastRequest().indices(index)));
    for (Tuple<ShardId, ActionListener<ReplicationResponse>> shardRequests : broadcastReplicationAction.capturedShardRequests) {
        if (randomBoolean()) {
            shardRequests.v2().onFailure(new NoShardAvailableActionException(shardRequests.v1()));
        } else {
            shardRequests.v2().onFailure(new UnavailableShardsException(shardRequests.v1(), "test exception"));
        }
    }
    response.get();
    logger.info("total shards: {}, ", response.get().getTotalShards());
    // we expect no failures here because UnavailableShardsException does not count as failed
    assertBroadcastResponse(2, 0, 0, response.get(), null);
}

Example 2 with NoShardAvailableActionException

use of org.elasticsearch.action.NoShardAvailableActionException in project elasticsearch by elastic.

the class ElasticsearchExceptionTests method testFailureToAndFromXContentWithDetails.

public void testFailureToAndFromXContentWithDetails() throws IOException {
    final XContent xContent = randomFrom(XContentType.values()).xContent();
    Exception failure;
    Throwable failureCause;
    ElasticsearchException expected;
    ElasticsearchException expectedCause;
    ElasticsearchException suppressed;
    switch(randomIntBetween(0, 6)) {
        case // Simple elasticsearch exception without cause
        0:
            failure = new NoNodeAvailableException("A");
            expected = new ElasticsearchException("Elasticsearch exception [type=no_node_available_exception, reason=A]");
            expected.addSuppressed(new ElasticsearchException("Elasticsearch exception [type=no_node_available_exception, reason=A]"));
            break;
        case // Simple elasticsearch exception with headers (other metadata of type number are not parsed)
        1:
            failure = new CircuitBreakingException("B", 5_000, 2_000);
            ((ElasticsearchException) failure).addHeader("header_name", "0", "1");
            expected = new ElasticsearchException("Elasticsearch exception [type=circuit_breaking_exception, reason=B]");
            expected.addHeader("header_name", "0", "1");
            suppressed = new ElasticsearchException("Elasticsearch exception [type=circuit_breaking_exception, reason=B]");
            suppressed.addHeader("header_name", "0", "1");
            expected.addSuppressed(suppressed);
            break;
        case // Elasticsearch exception with a cause, headers and parsable metadata
        2:
            failureCause = new NullPointerException("var is null");
            failure = new ScriptException("C", failureCause, singletonList("stack"), "test", "painless");
            ((ElasticsearchException) failure).addHeader("script_name", "my_script");
            expectedCause = new ElasticsearchException("Elasticsearch exception [type=null_pointer_exception, reason=var is null]");
            expected = new ElasticsearchException("Elasticsearch exception [type=script_exception, reason=C]", expectedCause);
            expected.addHeader("script_name", "my_script");
            expected.addMetadata("es.lang", "painless");
            expected.addMetadata("es.script", "test");
            expected.addMetadata("es.script_stack", "stack");
            suppressed = new ElasticsearchException("Elasticsearch exception [type=script_exception, reason=C]");
            suppressed.addHeader("script_name", "my_script");
            suppressed.addMetadata("es.lang", "painless");
            suppressed.addMetadata("es.script", "test");
            suppressed.addMetadata("es.script_stack", "stack");
            expected.addSuppressed(suppressed);
            break;
        case // JDK exception without cause
        3:
            failure = new IllegalStateException("D");
            expected = new ElasticsearchException("Elasticsearch exception [type=illegal_state_exception, reason=D]");
            suppressed = new ElasticsearchException("Elasticsearch exception [type=illegal_state_exception, reason=D]");
            expected.addSuppressed(suppressed);
            break;
        case // JDK exception with cause
        4:
            failureCause = new RoutingMissingException("idx", "type", "id");
            failure = new RuntimeException("E", failureCause);
            expectedCause = new ElasticsearchException("Elasticsearch exception [type=routing_missing_exception, " + "reason=routing is required for [idx]/[type]/[id]]");
            expectedCause.addMetadata("es.index", "idx");
            expectedCause.addMetadata("es.index_uuid", "_na_");
            expected = new ElasticsearchException("Elasticsearch exception [type=runtime_exception, reason=E]", expectedCause);
            suppressed = new ElasticsearchException("Elasticsearch exception [type=runtime_exception, reason=E]");
            expected.addSuppressed(suppressed);
            break;
        case // Wrapped exception with cause
        5:
            failureCause = new FileAlreadyExistsException("File exists");
            failure = new BroadcastShardOperationFailedException(new ShardId("_index", "_uuid", 5), "F", failureCause);
            expected = new ElasticsearchException("Elasticsearch exception [type=file_already_exists_exception, reason=File exists]");
            // strangely, the wrapped exception appears as the root cause...
            suppressed = new ElasticsearchException("Elasticsearch exception [type=broadcast_shard_operation_failed_exception, " + "reason=F]");
            expected.addSuppressed(suppressed);
            break;
        case // SearchPhaseExecutionException with cause and multiple failures
        6:
            DiscoveryNode node = new DiscoveryNode("node_g", buildNewFakeTransportAddress(), Version.CURRENT);
            failureCause = new NodeClosedException(node);
            failureCause = new NoShardAvailableActionException(new ShardId("_index_g", "_uuid_g", 6), "node_g", failureCause);
            ShardSearchFailure[] shardFailures = new ShardSearchFailure[] { new ShardSearchFailure(new ParsingException(0, 0, "Parsing g", null), new SearchShardTarget("node_g", new ShardId(new Index("_index_g", "_uuid_g"), 61))), new ShardSearchFailure(new RepositoryException("repository_g", "Repo"), new SearchShardTarget("node_g", new ShardId(new Index("_index_g", "_uuid_g"), 62))), new ShardSearchFailure(new SearchContextMissingException(0L), null) };
            failure = new SearchPhaseExecutionException("phase_g", "G", failureCause, shardFailures);
            expectedCause = new ElasticsearchException("Elasticsearch exception [type=node_closed_exception, " + "reason=node closed " + node + "]");
            expectedCause = new ElasticsearchException("Elasticsearch exception [type=no_shard_available_action_exception, " + "reason=node_g]", expectedCause);
            expectedCause.addMetadata("es.index", "_index_g");
            expectedCause.addMetadata("es.index_uuid", "_uuid_g");
            expectedCause.addMetadata("es.shard", "6");
            expected = new ElasticsearchException("Elasticsearch exception [type=search_phase_execution_exception, " + "reason=G]", expectedCause);
            expected.addMetadata("es.phase", "phase_g");
            expected.addSuppressed(new ElasticsearchException("Elasticsearch exception [type=parsing_exception, reason=Parsing g]"));
            expected.addSuppressed(new ElasticsearchException("Elasticsearch exception [type=repository_exception, " + "reason=[repository_g] Repo]"));
            expected.addSuppressed(new ElasticsearchException("Elasticsearch exception [type=search_context_missing_exception, " + "reason=No search context found for id [0]]"));
            break;
        default:
            throw new UnsupportedOperationException("Failed to generate randomized failure");
    }
    Exception finalFailure = failure;
    BytesReference failureBytes = XContentHelper.toXContent((builder, params) -> {
        ElasticsearchException.generateFailureXContent(builder, params, finalFailure, true);
        return builder;
    }, xContent.type(), randomBoolean());
    try (XContentParser parser = createParser(xContent, failureBytes)) {
        failureBytes = shuffleXContent(parser, randomBoolean()).bytes();
    }
    ElasticsearchException parsedFailure;
    try (XContentParser parser = createParser(xContent, failureBytes)) {
        assertEquals(XContentParser.Token.START_OBJECT, parser.nextToken());
        assertEquals(XContentParser.Token.FIELD_NAME, parser.nextToken());
        parsedFailure = ElasticsearchException.failureFromXContent(parser);
        assertEquals(XContentParser.Token.END_OBJECT, parser.nextToken());
        assertNull(parser.nextToken());
    }
    assertDeepEquals(expected, parsedFailure);
}

Example 3 with NoShardAvailableActionException

use of org.elasticsearch.action.NoShardAvailableActionException in project elasticsearch by elastic.

the class InitialSearchPhase method run.

@Override
public final void run() throws IOException {
    int shardIndex = -1;
    for (final ShardIterator shardIt : shardsIts) {
        shardIndex++;
        final ShardRouting shard = shardIt.nextOrNull();
        if (shard != null) {
            performPhaseOnShard(shardIndex, shardIt, shard);
        } else {
            // really, no shards active in this group
            onShardFailure(shardIndex, null, null, shardIt, new NoShardAvailableActionException(shardIt.shardId()));
        }
    }
}

Example 4 with NoShardAvailableActionException

use of org.elasticsearch.action.NoShardAvailableActionException in project elasticsearch by elastic.

the class DiscoveryWithServiceDisruptionsIT method testAckedIndexing.

/**
     * Test that we do not loose document whose indexing request was successful, under a randomly selected disruption scheme
     * We also collect & report the type of indexing failures that occur.
     * <p>
     * This test is a superset of tests run in the Jepsen test suite, with the exception of versioned updates
     */
@TestLogging("_root:DEBUG,org.elasticsearch.action.bulk:TRACE,org.elasticsearch.action.get:TRACE,discovery:TRACE," + "org.elasticsearch.cluster.service:TRACE,org.elasticsearch.indices.recovery:TRACE," + "org.elasticsearch.indices.cluster:TRACE,org.elasticsearch.index.shard:TRACE")
public void testAckedIndexing() throws Exception {
    final int seconds = !(TEST_NIGHTLY && rarely()) ? 1 : 5;
    final String timeout = seconds + "s";
    final List<String> nodes = startCluster(rarely() ? 5 : 3);
    assertAcked(prepareCreate("test").setSettings(Settings.builder().put(IndexMetaData.SETTING_NUMBER_OF_SHARDS, 1 + randomInt(2)).put(IndexMetaData.SETTING_NUMBER_OF_REPLICAS, randomInt(2)).put(IndexSettings.INDEX_SEQ_NO_CHECKPOINT_SYNC_INTERVAL.getKey(), randomBoolean() ? "5s" : "200ms")));
    ensureGreen();
    ServiceDisruptionScheme disruptionScheme = addRandomDisruptionScheme();
    logger.info("disruption scheme [{}] added", disruptionScheme);
    // id -> node sent.
    final ConcurrentHashMap<String, String> ackedDocs = new ConcurrentHashMap<>();
    final AtomicBoolean stop = new AtomicBoolean(false);
    List<Thread> indexers = new ArrayList<>(nodes.size());
    List<Semaphore> semaphores = new ArrayList<>(nodes.size());
    final AtomicInteger idGenerator = new AtomicInteger(0);
    final AtomicReference<CountDownLatch> countDownLatchRef = new AtomicReference<>();
    final List<Exception> exceptedExceptions = Collections.synchronizedList(new ArrayList<Exception>());
    logger.info("starting indexers");
    try {
        for (final String node : nodes) {
            final Semaphore semaphore = new Semaphore(0);
            semaphores.add(semaphore);
            final Client client = client(node);
            final String name = "indexer_" + indexers.size();
            final int numPrimaries = getNumShards("test").numPrimaries;
            Thread thread = new Thread(() -> {
                while (!stop.get()) {
                    String id = null;
                    try {
                        if (!semaphore.tryAcquire(10, TimeUnit.SECONDS)) {
                            continue;
                        }
                        logger.info("[{}] Acquired semaphore and it has {} permits left", name, semaphore.availablePermits());
                        try {
                            id = Integer.toString(idGenerator.incrementAndGet());
                            int shard = Math.floorMod(Murmur3HashFunction.hash(id), numPrimaries);
                            logger.trace("[{}] indexing id [{}] through node [{}] targeting shard [{}]", name, id, node, shard);
                            IndexResponse response = client.prepareIndex("test", "type", id).setSource("{}", XContentType.JSON).setTimeout(timeout).get(timeout);
                            assertEquals(DocWriteResponse.Result.CREATED, response.getResult());
                            ackedDocs.put(id, node);
                            logger.trace("[{}] indexed id [{}] through node [{}]", name, id, node);
                        } catch (ElasticsearchException e) {
                            exceptedExceptions.add(e);
                            final String docId = id;
                            logger.trace((Supplier<?>) () -> new ParameterizedMessage("[{}] failed id [{}] through node [{}]", name, docId, node), e);
                        } finally {
                            countDownLatchRef.get().countDown();
                            logger.trace("[{}] decreased counter : {}", name, countDownLatchRef.get().getCount());
                        }
                    } catch (InterruptedException e) {
                    // fine - semaphore interrupt
                    } catch (AssertionError | Exception e) {
                        logger.info((Supplier<?>) () -> new ParameterizedMessage("unexpected exception in background thread of [{}]", node), e);
                    }
                }
            });
            thread.setName(name);
            thread.start();
            indexers.add(thread);
        }
        int docsPerIndexer = randomInt(3);
        logger.info("indexing {} docs per indexer before partition", docsPerIndexer);
        countDownLatchRef.set(new CountDownLatch(docsPerIndexer * indexers.size()));
        for (Semaphore semaphore : semaphores) {
            semaphore.release(docsPerIndexer);
        }
        assertTrue(countDownLatchRef.get().await(1, TimeUnit.MINUTES));
        for (int iter = 1 + randomInt(2); iter > 0; iter--) {
            logger.info("starting disruptions & indexing (iteration [{}])", iter);
            disruptionScheme.startDisrupting();
            docsPerIndexer = 1 + randomInt(5);
            logger.info("indexing {} docs per indexer during partition", docsPerIndexer);
            countDownLatchRef.set(new CountDownLatch(docsPerIndexer * indexers.size()));
            Collections.shuffle(semaphores, random());
            for (Semaphore semaphore : semaphores) {
                assertThat(semaphore.availablePermits(), equalTo(0));
                semaphore.release(docsPerIndexer);
            }
            logger.info("waiting for indexing requests to complete");
            assertTrue(countDownLatchRef.get().await(docsPerIndexer * seconds * 1000 + 2000, TimeUnit.MILLISECONDS));
            logger.info("stopping disruption");
            disruptionScheme.stopDisrupting();
            for (String node : internalCluster().getNodeNames()) {
                ensureStableCluster(nodes.size(), TimeValue.timeValueMillis(disruptionScheme.expectedTimeToHeal().millis() + DISRUPTION_HEALING_OVERHEAD.millis()), true, node);
            }
            ensureGreen("test");
            logger.info("validating successful docs");
            assertBusy(() -> {
                for (String node : nodes) {
                    try {
                        logger.debug("validating through node [{}] ([{}] acked docs)", node, ackedDocs.size());
                        for (String id : ackedDocs.keySet()) {
                            assertTrue("doc [" + id + "] indexed via node [" + ackedDocs.get(id) + "] not found", client(node).prepareGet("test", "type", id).setPreference("_local").get().isExists());
                        }
                    } catch (AssertionError | NoShardAvailableActionException e) {
                        throw new AssertionError(e.getMessage() + " (checked via node [" + node + "]", e);
                    }
                }
            }, 30, TimeUnit.SECONDS);
            logger.info("done validating (iteration [{}])", iter);
        }
    } finally {
        if (exceptedExceptions.size() > 0) {
            StringBuilder sb = new StringBuilder();
            for (Exception e : exceptedExceptions) {
                sb.append("\n").append(e.getMessage());
            }
            logger.debug("Indexing exceptions during disruption: {}", sb);
        }
        logger.info("shutting down indexers");
        stop.set(true);
        for (Thread indexer : indexers) {
            indexer.interrupt();
            indexer.join(60000);
        }
    }
}

Example 5 with NoShardAvailableActionException

use of org.elasticsearch.action.NoShardAvailableActionException in project crate by crate.

the class ClusterDisruptionIT method testAckedIndexing.

/**
 * Test that we do not loose document whose indexing request was successful, under a randomly selected disruption scheme
 * We also collect & report the type of indexing failures that occur.
 * <p>
 * This test is a superset of tests run in the Jepsen test suite, with the exception of versioned updates
 */
@TestLogging("_root:DEBUG,org.elasticsearch.action.bulk:TRACE,org.elasticsearch.action.get:TRACE," + "org.elasticsearch.discovery:TRACE,org.elasticsearch.action.support.replication:TRACE," + "org.elasticsearch.cluster.service:TRACE,org.elasticsearch.indices.recovery:TRACE," + "org.elasticsearch.indices.cluster:TRACE,org.elasticsearch.index.shard:TRACE")
@Test
public void testAckedIndexing() throws Exception {
    final List<String> nodes = startCluster(3);
    int numberOfShards = 1 + randomInt(2);
    int replicas = randomInt(2);
    logger.info("creating table t clustered into {} shards with {} replicas", numberOfShards, replicas);
    execute("create table t (id int primary key, x string) clustered into " + numberOfShards + " shards " + "with (number_of_replicas = " + replicas + ", \"write.wait_for_active_shards\" = 1, \"global_checkpoint_sync.interval\"='1s')");
    ensureGreen();
    ServiceDisruptionScheme disruptionScheme = addRandomDisruptionScheme();
    logger.info("disruption scheme [{}] added", disruptionScheme);
    // id -> node sent.
    final ConcurrentHashMap<String, String> ackedDocs = new ConcurrentHashMap<>();
    final AtomicBoolean stop = new AtomicBoolean(false);
    List<Thread> indexers = new ArrayList<>(nodes.size());
    List<Semaphore> semaphores = new ArrayList<>(nodes.size());
    final AtomicInteger idGenerator = new AtomicInteger(0);
    final AtomicReference<CountDownLatch> countDownLatchRef = new AtomicReference<>();
    final List<Exception> exceptedExceptions = new CopyOnWriteArrayList<>();
    logger.info("starting indexers");
    try {
        for (final String node : nodes) {
            final Semaphore semaphore = new Semaphore(0);
            semaphores.add(semaphore);
            final String name = "indexer_" + indexers.size();
            Thread thread = new Thread(() -> {
                while (!stop.get()) {
                    String id = null;
                    try {
                        if (!semaphore.tryAcquire(10, TimeUnit.SECONDS)) {
                            continue;
                        }
                        logger.info("[{}] Acquired semaphore and it has {} permits left", name, semaphore.availablePermits());
                        try {
                            id = String.valueOf(idGenerator.incrementAndGet());
                            int shard = Math.floorMod(Murmur3HashFunction.hash(id), numberOfShards);
                            logger.trace("[{}] indexing id [{}] through node [{}] targeting shard [{}]", name, id, node, shard);
                            execute("insert into t (id, x) values (?, ?)", new Object[] { id, randomInt(100) }, node, TimeValue.timeValueSeconds(1L));
                            ackedDocs.put(id, node);
                            logger.trace("[{}] indexed id [{}] through node [{}], response [{}]", name, id, node, response);
                        } catch (ElasticsearchException | DuplicateKeyException e) {
                            exceptedExceptions.add(e);
                            final String rowId = id;
                            logger.trace(() -> new ParameterizedMessage("[{}] failed id [{}] through node [{}]", name, rowId, node), e);
                        } finally {
                            countDownLatchRef.get().countDown();
                            logger.trace("[{}] decreased counter : {}", name, countDownLatchRef.get().getCount());
                        }
                    } catch (InterruptedException e) {
                    // fine - semaphore interrupt
                    } catch (AssertionError | Exception e) {
                        logger.trace(() -> new ParameterizedMessage("unexpected exception in background thread of [{}]", node), e);
                    }
                }
            });
            thread.setName(name);
            thread.start();
            indexers.add(thread);
        }
        int docsPerIndexer = randomInt(3);
        logger.info("indexing {} docs per indexer before partition", docsPerIndexer);
        countDownLatchRef.set(new CountDownLatch(docsPerIndexer * indexers.size()));
        for (Semaphore semaphore : semaphores) {
            semaphore.release(docsPerIndexer);
        }
        assertTrue(countDownLatchRef.get().await(1, TimeUnit.MINUTES));
        for (int iter = 1 + randomInt(1); iter > 0; iter--) {
            logger.info("starting disruptions & indexing (iteration [{}])", iter);
            disruptionScheme.startDisrupting();
            docsPerIndexer = randomIntBetween(1, 4);
            logger.info("indexing {} docs per indexer during partition", docsPerIndexer);
            countDownLatchRef.set(new CountDownLatch(docsPerIndexer * indexers.size()));
            Collections.shuffle(semaphores, random());
            for (Semaphore semaphore : semaphores) {
                assertThat(semaphore.availablePermits(), equalTo(0));
                semaphore.release(docsPerIndexer);
            }
            logger.info("waiting for indexing requests to complete");
            assertThat("indexing requests must complete", countDownLatchRef.get().await(20, TimeUnit.SECONDS), is(true));
            logger.info("stopping disruption");
            disruptionScheme.stopDisrupting();
            for (String node : internalCluster().getNodeNames()) {
                ensureStableCluster(nodes.size(), TimeValue.timeValueMillis(disruptionScheme.expectedTimeToHeal().millis() + DISRUPTION_HEALING_OVERHEAD.millis()), true, node);
            }
            // is the super-connected node and recovery source and target are on opposite sides of the bridge
            if (disruptionScheme instanceof NetworkDisruption && ((NetworkDisruption) disruptionScheme).getDisruptedLinks() instanceof Bridge) {
                logger.warn("retrying failed allocations in case of a bridge partition");
                execute("ALTER CLUSTER REROUTE RETRY FAILED");
            }
            ensureGreen();
            logger.info("validating successful docs");
            assertBusy(() -> {
                for (String node : nodes) {
                    try {
                        logger.debug("validating through node [{}] ([{}] acked docs)", node, ackedDocs.size());
                        for (String id : ackedDocs.keySet()) {
                            execute("select * from t where id = ?", new Object[] { id }, node);
                            assertThat("doc [" + id + "] indexed via node [" + ackedDocs.get(id) + "] not found", response.rowCount(), is(1L));
                        }
                    } catch (AssertionError | NoShardAvailableActionException e) {
                        throw new AssertionError(e.getMessage() + " (checked via node [" + node + "]", e);
                    }
                }
            }, 30, TimeUnit.SECONDS);
            logger.info("done validating (iteration [{}])", iter);
        }
    } finally {
        logger.info("shutting down indexers");
        stop.set(true);
        for (Thread indexer : indexers) {
            indexer.interrupt();
            indexer.join(60000);
        }
        if (exceptedExceptions.size() > 0) {
            StringBuilder sb = new StringBuilder();
            for (Exception e : exceptedExceptions) {
                sb.append("\n").append(e.getMessage());
            }
            logger.debug("Indexing exceptions during disruption: {}", sb);
        }
    }
}