Examples with RecoveryFailedException org.opensearch.indices.recovery.RecoveryFailedException used on opensource projects

Example 1 with RecoveryFailedException

use of org.opensearch.indices.recovery.RecoveryFailedException in project OpenSearch by opensearch-project.

the class IndexShard method startRecovery.

public void startRecovery(RecoveryState recoveryState, PeerRecoveryTargetService recoveryTargetService, PeerRecoveryTargetService.RecoveryListener recoveryListener, RepositoriesService repositoriesService, Consumer<MappingMetadata> mappingUpdateConsumer, IndicesService indicesService) {
    // }
    assert recoveryState.getRecoverySource().equals(shardRouting.recoverySource());
    switch(recoveryState.getRecoverySource().getType()) {
        case EMPTY_STORE:
        case EXISTING_STORE:
            executeRecovery("from store", recoveryState, recoveryListener, this::recoverFromStore);
            break;
        case PEER:
            try {
                markAsRecovering("from " + recoveryState.getSourceNode(), recoveryState);
                recoveryTargetService.startRecovery(this, recoveryState.getSourceNode(), recoveryListener);
            } catch (Exception e) {
                failShard("corrupted preexisting index", e);
                recoveryListener.onRecoveryFailure(recoveryState, new RecoveryFailedException(recoveryState, null, e), true);
            }
            break;
        case SNAPSHOT:
            final String repo = ((SnapshotRecoverySource) recoveryState.getRecoverySource()).snapshot().getRepository();
            executeRecovery("from snapshot", recoveryState, recoveryListener, l -> restoreFromRepository(repositoriesService.repository(repo), l));
            break;
        case LOCAL_SHARDS:
            final IndexMetadata indexMetadata = indexSettings().getIndexMetadata();
            final Index resizeSourceIndex = indexMetadata.getResizeSourceIndex();
            final List<IndexShard> startedShards = new ArrayList<>();
            final IndexService sourceIndexService = indicesService.indexService(resizeSourceIndex);
            final Set<ShardId> requiredShards;
            final int numShards;
            if (sourceIndexService != null) {
                requiredShards = IndexMetadata.selectRecoverFromShards(shardId().id(), sourceIndexService.getMetadata(), indexMetadata.getNumberOfShards());
                for (IndexShard shard : sourceIndexService) {
                    if (shard.state() == IndexShardState.STARTED && requiredShards.contains(shard.shardId())) {
                        startedShards.add(shard);
                    }
                }
                numShards = requiredShards.size();
            } else {
                numShards = -1;
                requiredShards = Collections.emptySet();
            }
            if (numShards == startedShards.size()) {
                assert requiredShards.isEmpty() == false;
                executeRecovery("from local shards", recoveryState, recoveryListener, l -> recoverFromLocalShards(mappingUpdateConsumer, startedShards.stream().filter((s) -> requiredShards.contains(s.shardId())).collect(Collectors.toList()), l));
            } else {
                final RuntimeException e;
                if (numShards == -1) {
                    e = new IndexNotFoundException(resizeSourceIndex);
                } else {
                    e = new IllegalStateException("not all required shards of index " + resizeSourceIndex + " are started yet, expected " + numShards + " found " + startedShards.size() + " can't recover shard " + shardId());
                }
                throw e;
            }
            break;
        default:
            throw new IllegalArgumentException("Unknown recovery source " + recoveryState.getRecoverySource());
    }
}

Example 2 with RecoveryFailedException

use of org.opensearch.indices.recovery.RecoveryFailedException in project OpenSearch by opensearch-project.

the class RecoveriesCollectionTests method testRecoveryTimeout.

public void testRecoveryTimeout() throws Exception {
    try (ReplicationGroup shards = createGroup(0)) {
        final RecoveriesCollection collection = new RecoveriesCollection(logger, threadPool);
        final AtomicBoolean failed = new AtomicBoolean();
        final CountDownLatch latch = new CountDownLatch(1);
        final long recoveryId = startRecovery(collection, shards.getPrimaryNode(), shards.addReplica(), new PeerRecoveryTargetService.RecoveryListener() {

            @Override
            public void onRecoveryDone(RecoveryState state) {
                latch.countDown();
            }

            @Override
            public void onRecoveryFailure(RecoveryState state, RecoveryFailedException e, boolean sendShardFailure) {
                failed.set(true);
                latch.countDown();
            }
        }, TimeValue.timeValueMillis(100));
        try {
            latch.await(30, TimeUnit.SECONDS);
            assertTrue("recovery failed to timeout", failed.get());
        } finally {
            collection.cancelRecovery(recoveryId, "meh");
        }
    }
}

Example 3 with RecoveryFailedException

use of org.opensearch.indices.recovery.RecoveryFailedException in project OpenSearch by opensearch-project.

the class IndexShardTestCase method recoverUnstartedReplica.

/**
 * Recovers a replica from the give primary, allow the user to supply a custom recovery target. A typical usage of a custom recovery
 * target is to assert things in the various stages of recovery.
 *
 * Note: this method keeps the shard in {@link IndexShardState#POST_RECOVERY} and doesn't start it.
 *
 * @param replica                the recovery target shard
 * @param primary                the recovery source shard
 * @param targetSupplier         supplies an instance of {@link RecoveryTarget}
 * @param markAsRecovering       set to {@code false} if the replica is marked as recovering
 */
protected final void recoverUnstartedReplica(final IndexShard replica, final IndexShard primary, final BiFunction<IndexShard, DiscoveryNode, RecoveryTarget> targetSupplier, final boolean markAsRecovering, final Set<String> inSyncIds, final IndexShardRoutingTable routingTable) throws IOException {
    final DiscoveryNode pNode = getFakeDiscoNode(primary.routingEntry().currentNodeId());
    final DiscoveryNode rNode = getFakeDiscoNode(replica.routingEntry().currentNodeId());
    if (markAsRecovering) {
        replica.markAsRecovering("remote", new RecoveryState(replica.routingEntry(), pNode, rNode));
    } else {
        assertEquals(replica.state(), IndexShardState.RECOVERING);
    }
    replica.prepareForIndexRecovery();
    final RecoveryTarget recoveryTarget = targetSupplier.apply(replica, pNode);
    final long startingSeqNo = recoveryTarget.indexShard().recoverLocallyUpToGlobalCheckpoint();
    final StartRecoveryRequest request = PeerRecoveryTargetService.getStartRecoveryRequest(logger, rNode, recoveryTarget, startingSeqNo);
    int fileChunkSizeInBytes = Math.toIntExact(randomBoolean() ? RecoverySettings.DEFAULT_CHUNK_SIZE.getBytes() : randomIntBetween(1, 10 * 1024 * 1024));
    final RecoverySourceHandler recovery = new RecoverySourceHandler(primary, new AsyncRecoveryTarget(recoveryTarget, threadPool.generic()), threadPool, request, fileChunkSizeInBytes, between(1, 8), between(1, 8));
    primary.updateShardState(primary.routingEntry(), primary.getPendingPrimaryTerm(), null, currentClusterStateVersion.incrementAndGet(), inSyncIds, routingTable);
    try {
        PlainActionFuture<RecoveryResponse> future = new PlainActionFuture<>();
        recovery.recoverToTarget(future);
        future.actionGet();
        recoveryTarget.markAsDone();
    } catch (Exception e) {
        recoveryTarget.fail(new RecoveryFailedException(request, e), false);
        throw e;
    }
}