Examples with CorruptIndexException org.apache.lucene.index.CorruptIndexException used on opensource projects

Example 1 with CorruptIndexException

use of org.apache.lucene.index.CorruptIndexException in project elasticsearch by elastic.

the class RecoverySourceHandler method phase1.

/**
     * Perform phase1 of the recovery operations. Once this {@link IndexCommit}
     * snapshot has been performed no commit operations (files being fsync'd)
     * are effectively allowed on this index until all recovery phases are done
     * <p>
     * Phase1 examines the segment files on the target node and copies over the
     * segments that are missing. Only segments that have the same size and
     * checksum can be reused
     */
public void phase1(final IndexCommit snapshot, final Translog.View translogView) {
    cancellableThreads.checkForCancel();
    // Total size of segment files that are recovered
    long totalSize = 0;
    // Total size of segment files that were able to be re-used
    long existingTotalSize = 0;
    final Store store = shard.store();
    store.incRef();
    try {
        StopWatch stopWatch = new StopWatch().start();
        final Store.MetadataSnapshot recoverySourceMetadata;
        try {
            recoverySourceMetadata = store.getMetadata(snapshot);
        } catch (CorruptIndexException | IndexFormatTooOldException | IndexFormatTooNewException ex) {
            shard.failShard("recovery", ex);
            throw ex;
        }
        for (String name : snapshot.getFileNames()) {
            final StoreFileMetaData md = recoverySourceMetadata.get(name);
            if (md == null) {
                logger.info("Snapshot differs from actual index for file: {} meta: {}", name, recoverySourceMetadata.asMap());
                throw new CorruptIndexException("Snapshot differs from actual index - maybe index was removed metadata has " + recoverySourceMetadata.asMap().size() + " files", name);
            }
        }
        // Generate a "diff" of all the identical, different, and missing
        // segment files on the target node, using the existing files on
        // the source node
        String recoverySourceSyncId = recoverySourceMetadata.getSyncId();
        String recoveryTargetSyncId = request.metadataSnapshot().getSyncId();
        final boolean recoverWithSyncId = recoverySourceSyncId != null && recoverySourceSyncId.equals(recoveryTargetSyncId);
        if (recoverWithSyncId) {
            final long numDocsTarget = request.metadataSnapshot().getNumDocs();
            final long numDocsSource = recoverySourceMetadata.getNumDocs();
            if (numDocsTarget != numDocsSource) {
                throw new IllegalStateException("try to recover " + request.shardId() + " from primary shard with sync id but number " + "of docs differ: " + numDocsSource + " (" + request.sourceNode().getName() + ", primary) vs " + numDocsTarget + "(" + request.targetNode().getName() + ")");
            }
            // we shortcut recovery here because we have nothing to copy. but we must still start the engine on the target.
            // so we don't return here
            logger.trace("skipping [phase1]- identical sync id [{}] found on both source and target", recoverySourceSyncId);
        } else {
            final Store.RecoveryDiff diff = recoverySourceMetadata.recoveryDiff(request.metadataSnapshot());
            for (StoreFileMetaData md : diff.identical) {
                response.phase1ExistingFileNames.add(md.name());
                response.phase1ExistingFileSizes.add(md.length());
                existingTotalSize += md.length();
                if (logger.isTraceEnabled()) {
                    logger.trace("recovery [phase1]: not recovering [{}], exist in local store and has checksum [{}]," + " size [{}]", md.name(), md.checksum(), md.length());
                }
                totalSize += md.length();
            }
            List<StoreFileMetaData> phase1Files = new ArrayList<>(diff.different.size() + diff.missing.size());
            phase1Files.addAll(diff.different);
            phase1Files.addAll(diff.missing);
            for (StoreFileMetaData md : phase1Files) {
                if (request.metadataSnapshot().asMap().containsKey(md.name())) {
                    logger.trace("recovery [phase1]: recovering [{}], exists in local store, but is different: remote [{}], local [{}]", md.name(), request.metadataSnapshot().asMap().get(md.name()), md);
                } else {
                    logger.trace("recovery [phase1]: recovering [{}], does not exist in remote", md.name());
                }
                response.phase1FileNames.add(md.name());
                response.phase1FileSizes.add(md.length());
                totalSize += md.length();
            }
            response.phase1TotalSize = totalSize;
            response.phase1ExistingTotalSize = existingTotalSize;
            logger.trace("recovery [phase1]: recovering_files [{}] with total_size [{}], reusing_files [{}] with total_size [{}]", response.phase1FileNames.size(), new ByteSizeValue(totalSize), response.phase1ExistingFileNames.size(), new ByteSizeValue(existingTotalSize));
            cancellableThreads.execute(() -> recoveryTarget.receiveFileInfo(response.phase1FileNames, response.phase1FileSizes, response.phase1ExistingFileNames, response.phase1ExistingFileSizes, translogView.totalOperations()));
            // How many bytes we've copied since we last called RateLimiter.pause
            final Function<StoreFileMetaData, OutputStream> outputStreamFactories = md -> new BufferedOutputStream(new RecoveryOutputStream(md, translogView), chunkSizeInBytes);
            sendFiles(store, phase1Files.toArray(new StoreFileMetaData[phase1Files.size()]), outputStreamFactories);
            // are deleted
            try {
                cancellableThreads.executeIO(() -> recoveryTarget.cleanFiles(translogView.totalOperations(), recoverySourceMetadata));
            } catch (RemoteTransportException | IOException targetException) {
                final IOException corruptIndexException;
                //   - maybe due to old segments without checksums or length only checks
                if ((corruptIndexException = ExceptionsHelper.unwrapCorruption(targetException)) != null) {
                    try {
                        final Store.MetadataSnapshot recoverySourceMetadata1 = store.getMetadata(snapshot);
                        StoreFileMetaData[] metadata = StreamSupport.stream(recoverySourceMetadata1.spliterator(), false).toArray(size -> new StoreFileMetaData[size]);
                        ArrayUtil.timSort(metadata, (o1, o2) -> {
                            // check small files first
                            return Long.compare(o1.length(), o2.length());
                        });
                        for (StoreFileMetaData md : metadata) {
                            cancellableThreads.checkForCancel();
                            logger.debug("checking integrity for file {} after remove corruption exception", md);
                            if (store.checkIntegrityNoException(md) == false) {
                                // we are corrupted on the primary -- fail!
                                shard.failShard("recovery", corruptIndexException);
                                logger.warn("Corrupted file detected {} checksum mismatch", md);
                                throw corruptIndexException;
                            }
                        }
                    } catch (IOException ex) {
                        targetException.addSuppressed(ex);
                        throw targetException;
                    }
                    // corruption has happened on the way to replica
                    RemoteTransportException exception = new RemoteTransportException("File corruption occurred on recovery but " + "checksums are ok", null);
                    exception.addSuppressed(targetException);
                    logger.warn((org.apache.logging.log4j.util.Supplier<?>) () -> new ParameterizedMessage("{} Remote file corruption during finalization of recovery on node {}. local checksum OK", shard.shardId(), request.targetNode()), corruptIndexException);
                    throw exception;
                } else {
                    throw targetException;
                }
            }
        }
        logger.trace("recovery [phase1]: took [{}]", stopWatch.totalTime());
        response.phase1Time = stopWatch.totalTime().millis();
    } catch (Exception e) {
        throw new RecoverFilesRecoveryException(request.shardId(), response.phase1FileNames.size(), new ByteSizeValue(totalSize), e);
    } finally {
        store.decRef();
    }
}

Example 2 with CorruptIndexException

use of org.apache.lucene.index.CorruptIndexException in project elasticsearch by elastic.

the class RecoveryTarget method cleanFiles.

@Override
public void cleanFiles(int totalTranslogOps, Store.MetadataSnapshot sourceMetaData) throws IOException {
    state().getTranslog().totalOperations(totalTranslogOps);
    // first, we go and move files that were created with the recovery id suffix to
    // the actual names, its ok if we have a corrupted index here, since we have replicas
    // to recover from in case of a full cluster shutdown just when this code executes...
    renameAllTempFiles();
    final Store store = store();
    try {
        store.cleanupAndVerify("recovery CleanFilesRequestHandler", sourceMetaData);
    } catch (CorruptIndexException | IndexFormatTooNewException | IndexFormatTooOldException ex) {
        // its content on disk if possible.
        try {
            try {
                store.removeCorruptionMarker();
            } finally {
                // clean up and delete all files
                Lucene.cleanLuceneIndex(store.directory());
            }
        } catch (Exception e) {
            logger.debug("Failed to clean lucene index", e);
            ex.addSuppressed(e);
        }
        RecoveryFailedException rfe = new RecoveryFailedException(state(), "failed to clean after recovery", ex);
        fail(rfe, true);
        throw rfe;
    } catch (Exception ex) {
        RecoveryFailedException rfe = new RecoveryFailedException(state(), "failed to clean after recovery", ex);
        fail(rfe, true);
        throw rfe;
    }
}

Example 3 with CorruptIndexException

use of org.apache.lucene.index.CorruptIndexException in project elasticsearch by elastic.

the class ReplicationOperationTests method testReplication.

public void testReplication() throws Exception {
    final String index = "test";
    final ShardId shardId = new ShardId(index, "_na_", 0);
    ClusterState state = stateWithActivePrimary(index, true, randomInt(5));
    IndexMetaData indexMetaData = state.getMetaData().index(index);
    final long primaryTerm = indexMetaData.primaryTerm(0);
    final IndexShardRoutingTable indexShardRoutingTable = state.getRoutingTable().shardRoutingTable(shardId);
    ShardRouting primaryShard = indexShardRoutingTable.primaryShard();
    if (primaryShard.relocating() && randomBoolean()) {
        // simulate execution of the replication phase on the relocation target node after relocation source was marked as relocated
        state = ClusterState.builder(state).nodes(DiscoveryNodes.builder(state.nodes()).localNodeId(primaryShard.relocatingNodeId())).build();
        primaryShard = primaryShard.getTargetRelocatingShard();
    }
    // add a few in-sync allocation ids that don't have corresponding routing entries
    Set<String> staleAllocationIds = Sets.newHashSet(generateRandomStringArray(4, 10, false));
    state = ClusterState.builder(state).metaData(MetaData.builder(state.metaData()).put(IndexMetaData.builder(indexMetaData).putInSyncAllocationIds(0, Sets.union(indexMetaData.inSyncAllocationIds(0), staleAllocationIds)))).build();
    final Set<ShardRouting> expectedReplicas = getExpectedReplicas(shardId, state);
    final Map<ShardRouting, Exception> expectedFailures = new HashMap<>();
    final Set<ShardRouting> expectedFailedShards = new HashSet<>();
    for (ShardRouting replica : expectedReplicas) {
        if (randomBoolean()) {
            Exception t;
            boolean criticalFailure = randomBoolean();
            if (criticalFailure) {
                t = new CorruptIndexException("simulated", (String) null);
            } else {
                t = new IndexShardNotStartedException(shardId, IndexShardState.RECOVERING);
            }
            logger.debug("--> simulating failure on {} with [{}]", replica, t.getClass().getSimpleName());
            expectedFailures.put(replica, t);
            if (criticalFailure) {
                expectedFailedShards.add(replica);
            }
        }
    }
    Request request = new Request(shardId);
    PlainActionFuture<TestPrimary.Result> listener = new PlainActionFuture<>();
    final ClusterState finalState = state;
    final TestReplicaProxy replicasProxy = new TestReplicaProxy(expectedFailures);
    final TestPrimary primary = new TestPrimary(primaryShard, primaryTerm);
    final TestReplicationOperation op = new TestReplicationOperation(request, primary, listener, replicasProxy, () -> finalState);
    op.execute();
    assertThat(request.primaryTerm(), equalTo(primaryTerm));
    assertThat("request was not processed on primary", request.processedOnPrimary.get(), equalTo(true));
    assertThat(request.processedOnReplicas, equalTo(expectedReplicas));
    assertThat(replicasProxy.failedReplicas, equalTo(expectedFailedShards));
    assertThat(replicasProxy.markedAsStaleCopies, equalTo(staleAllocationIds));
    assertTrue("listener is not marked as done", listener.isDone());
    ShardInfo shardInfo = listener.actionGet().getShardInfo();
    assertThat(shardInfo.getFailed(), equalTo(expectedFailedShards.size()));
    assertThat(shardInfo.getFailures(), arrayWithSize(expectedFailedShards.size()));
    assertThat(shardInfo.getSuccessful(), equalTo(1 + expectedReplicas.size() - expectedFailures.size()));
    final List<ShardRouting> unassignedShards = indexShardRoutingTable.shardsWithState(ShardRoutingState.UNASSIGNED);
    final int totalShards = 1 + expectedReplicas.size() + unassignedShards.size();
    assertThat(shardInfo.getTotal(), equalTo(totalShards));
    assertThat(primary.knownLocalCheckpoints.remove(primaryShard.allocationId().getId()), equalTo(primary.localCheckpoint));
    assertThat(primary.knownLocalCheckpoints, equalTo(replicasProxy.generatedLocalCheckpoints));
}

Example 4 with CorruptIndexException

use of org.apache.lucene.index.CorruptIndexException in project elasticsearch by elastic.

the class ReplicationOperationTests method testDemotedPrimary.

public void testDemotedPrimary() throws Exception {
    final String index = "test";
    final ShardId shardId = new ShardId(index, "_na_", 0);
    ClusterState state = stateWithActivePrimary(index, true, 1 + randomInt(2), randomInt(2));
    IndexMetaData indexMetaData = state.getMetaData().index(index);
    final long primaryTerm = indexMetaData.primaryTerm(0);
    ShardRouting primaryShard = state.getRoutingTable().shardRoutingTable(shardId).primaryShard();
    if (primaryShard.relocating() && randomBoolean()) {
        // simulate execution of the replication phase on the relocation target node after relocation source was marked as relocated
        state = ClusterState.builder(state).nodes(DiscoveryNodes.builder(state.nodes()).localNodeId(primaryShard.relocatingNodeId())).build();
        primaryShard = primaryShard.getTargetRelocatingShard();
    }
    // add in-sync allocation id that doesn't have a corresponding routing entry
    state = ClusterState.builder(state).metaData(MetaData.builder(state.metaData()).put(IndexMetaData.builder(indexMetaData).putInSyncAllocationIds(0, Sets.union(indexMetaData.inSyncAllocationIds(0), Sets.newHashSet(randomAsciiOfLength(10)))))).build();
    final Set<ShardRouting> expectedReplicas = getExpectedReplicas(shardId, state);
    final Map<ShardRouting, Exception> expectedFailures = new HashMap<>();
    final ShardRouting failedReplica = randomFrom(new ArrayList<>(expectedReplicas));
    expectedFailures.put(failedReplica, new CorruptIndexException("simulated", (String) null));
    Request request = new Request(shardId);
    PlainActionFuture<TestPrimary.Result> listener = new PlainActionFuture<>();
    final ClusterState finalState = state;
    final boolean testPrimaryDemotedOnStaleShardCopies = randomBoolean();
    final TestReplicaProxy replicasProxy = new TestReplicaProxy(expectedFailures) {

        @Override
        public void failShardIfNeeded(ShardRouting replica, long primaryTerm, String message, Exception exception, Runnable onSuccess, Consumer<Exception> onPrimaryDemoted, Consumer<Exception> onIgnoredFailure) {
            if (testPrimaryDemotedOnStaleShardCopies) {
                super.failShardIfNeeded(replica, primaryTerm, message, exception, onSuccess, onPrimaryDemoted, onIgnoredFailure);
            } else {
                assertThat(replica, equalTo(failedReplica));
                onPrimaryDemoted.accept(new ElasticsearchException("the king is dead"));
            }
        }

        @Override
        public void markShardCopyAsStaleIfNeeded(ShardId shardId, String allocationId, long primaryTerm, Runnable onSuccess, Consumer<Exception> onPrimaryDemoted, Consumer<Exception> onIgnoredFailure) {
            if (testPrimaryDemotedOnStaleShardCopies) {
                onPrimaryDemoted.accept(new ElasticsearchException("the king is dead"));
            } else {
                super.markShardCopyAsStaleIfNeeded(shardId, allocationId, primaryTerm, onSuccess, onPrimaryDemoted, onIgnoredFailure);
            }
        }
    };
    AtomicBoolean primaryFailed = new AtomicBoolean();
    final TestPrimary primary = new TestPrimary(primaryShard, primaryTerm) {

        @Override
        public void failShard(String message, Exception exception) {
            assertTrue(primaryFailed.compareAndSet(false, true));
        }
    };
    final TestReplicationOperation op = new TestReplicationOperation(request, primary, listener, replicasProxy, () -> finalState);
    op.execute();
    assertThat("request was not processed on primary", request.processedOnPrimary.get(), equalTo(true));
    assertTrue("listener is not marked as done", listener.isDone());
    assertTrue(primaryFailed.get());
    assertListenerThrows("should throw exception to trigger retry", listener, ReplicationOperation.RetryOnPrimaryException.class);
}

Example 5 with CorruptIndexException

use of org.apache.lucene.index.CorruptIndexException in project elasticsearch by elastic.

the class RecoverySourceHandlerTests method testHandleExceptinoOnSendSendFiles.

public void testHandleExceptinoOnSendSendFiles() throws Throwable {
    Settings settings = Settings.builder().put("indices.recovery.concurrent_streams", 1).put("indices.recovery.concurrent_small_file_streams", 1).build();
    final RecoverySettings recoverySettings = new RecoverySettings(settings, service);
    final StartRecoveryRequest request = new StartRecoveryRequest(shardId, new DiscoveryNode("b", buildNewFakeTransportAddress(), emptyMap(), emptySet(), Version.CURRENT), new DiscoveryNode("b", buildNewFakeTransportAddress(), emptyMap(), emptySet(), Version.CURRENT), null, randomBoolean(), randomNonNegativeLong(), randomBoolean() ? SequenceNumbersService.UNASSIGNED_SEQ_NO : 0L);
    Path tempDir = createTempDir();
    Store store = newStore(tempDir, false);
    AtomicBoolean failedEngine = new AtomicBoolean(false);
    RecoverySourceHandler handler = new RecoverySourceHandler(null, null, request, () -> 0L, e -> () -> {
    }, recoverySettings.getChunkSize().bytesAsInt(), Settings.EMPTY) {

        @Override
        protected void failEngine(IOException cause) {
            assertFalse(failedEngine.get());
            failedEngine.set(true);
        }
    };
    Directory dir = store.directory();
    RandomIndexWriter writer = new RandomIndexWriter(random(), dir, newIndexWriterConfig());
    int numDocs = randomIntBetween(10, 100);
    for (int i = 0; i < numDocs; i++) {
        Document document = new Document();
        document.add(new StringField("id", Integer.toString(i), Field.Store.YES));
        document.add(newField("field", randomUnicodeOfCodepointLengthBetween(1, 10), TextField.TYPE_STORED));
        writer.addDocument(document);
    }
    writer.commit();
    writer.close();
    Store.MetadataSnapshot metadata = store.getMetadata(null);
    List<StoreFileMetaData> metas = new ArrayList<>();
    for (StoreFileMetaData md : metadata) {
        metas.add(md);
    }
    final boolean throwCorruptedIndexException = randomBoolean();
    Store targetStore = newStore(createTempDir(), false);
    try {
        handler.sendFiles(store, metas.toArray(new StoreFileMetaData[0]), (md) -> {
            if (throwCorruptedIndexException) {
                throw new RuntimeException(new CorruptIndexException("foo", "bar"));
            } else {
                throw new RuntimeException("boom");
            }
        });
        fail("exception index");
    } catch (RuntimeException ex) {
        assertNull(ExceptionsHelper.unwrapCorruption(ex));
        if (throwCorruptedIndexException) {
            assertEquals(ex.getMessage(), "[File corruption occurred on recovery but checksums are ok]");
        } else {
            assertEquals(ex.getMessage(), "boom");
        }
    } catch (CorruptIndexException ex) {
        fail("not expected here");
    }
    assertFalse(failedEngine.get());
    IOUtils.close(store, targetStore);
}