Search in sources :

Example 1 with BatchingNeoStores

use of org.neo4j.unsafe.impl.batchimport.store.BatchingNeoStores in project neo4j by neo4j.

the class ParallelBatchImporter method doImport.

@Override
public void doImport(Input input) throws IOException {
    log.info("Import starting");
    // Things that we need to close later. The reason they're not in the try-with-resource statement
    // is that we need to close, and set to null, at specific points preferably. So use good ol' finally block.
    NodeRelationshipCache nodeRelationshipCache = null;
    NodeLabelsCache nodeLabelsCache = null;
    long startTime = currentTimeMillis();
    CountingStoreUpdateMonitor storeUpdateMonitor = new CountingStoreUpdateMonitor();
    try (BatchingNeoStores neoStore = getBatchingNeoStores();
        CountsAccessor.Updater countsUpdater = neoStore.getCountsStore().reset(neoStore.getLastCommittedTransactionId());
        InputCache inputCache = new InputCache(fileSystem, storeDir, recordFormats, config)) {
        Collector badCollector = input.badCollector();
        // Some temporary caches and indexes in the import
        IoMonitor writeMonitor = new IoMonitor(neoStore.getIoTracer());
        IdMapper idMapper = input.idMapper();
        IdGenerator idGenerator = input.idGenerator();
        nodeRelationshipCache = new NodeRelationshipCache(AUTO, config.denseNodeThreshold());
        StatsProvider memoryUsageStats = new MemoryUsageStatsProvider(nodeRelationshipCache, idMapper);
        InputIterable<InputNode> nodes = input.nodes();
        InputIterable<InputRelationship> relationships = input.relationships();
        InputIterable<InputNode> cachedNodes = cachedForSure(nodes, inputCache.nodes(MAIN, true));
        InputIterable<InputRelationship> cachedRelationships = cachedForSure(relationships, inputCache.relationships(MAIN, true));
        RelationshipStore relationshipStore = neoStore.getRelationshipStore();
        // Stage 1 -- nodes, properties, labels
        NodeStage nodeStage = new NodeStage(config, writeMonitor, nodes, idMapper, idGenerator, neoStore, inputCache, neoStore.getLabelScanStore(), storeUpdateMonitor, nodeRelationshipCache, memoryUsageStats);
        executeStage(nodeStage);
        if (idMapper.needsPreparation()) {
            executeStage(new IdMapperPreparationStage(config, idMapper, cachedNodes, badCollector, memoryUsageStats));
            PrimitiveLongIterator duplicateNodeIds = badCollector.leftOverDuplicateNodesIds();
            if (duplicateNodeIds.hasNext()) {
                executeStage(new DeleteDuplicateNodesStage(config, duplicateNodeIds, neoStore));
            }
        }
        // Stage 2 -- calculate dense node threshold
        CalculateDenseNodesStage calculateDenseNodesStage = new CalculateDenseNodesStage(withBatchSize(config, config.batchSize() * 10), relationships, nodeRelationshipCache, idMapper, badCollector, inputCache, neoStore);
        executeStage(calculateDenseNodesStage);
        importRelationships(nodeRelationshipCache, storeUpdateMonitor, neoStore, writeMonitor, idMapper, cachedRelationships, inputCache, calculateDenseNodesStage.getRelationshipTypes(Long.MAX_VALUE), calculateDenseNodesStage.getRelationshipTypes(100));
        // Release this potentially really big piece of cached data
        long peakMemoryUsage = totalMemoryUsageOf(idMapper, nodeRelationshipCache);
        long highNodeId = nodeRelationshipCache.getHighNodeId();
        idMapper.close();
        idMapper = null;
        nodeRelationshipCache.close();
        nodeRelationshipCache = null;
        new RelationshipGroupDefragmenter(config, executionMonitor).run(max(max(peakMemoryUsage, highNodeId * 4), mebiBytes(1)), neoStore, highNodeId);
        // Stage 6 -- count nodes per label and labels per node
        nodeLabelsCache = new NodeLabelsCache(AUTO, neoStore.getLabelRepository().getHighId());
        memoryUsageStats = new MemoryUsageStatsProvider(nodeLabelsCache);
        executeStage(new NodeCountsStage(config, nodeLabelsCache, neoStore.getNodeStore(), neoStore.getLabelRepository().getHighId(), countsUpdater, memoryUsageStats));
        // Stage 7 -- count label-[type]->label
        executeStage(new RelationshipCountsStage(config, nodeLabelsCache, relationshipStore, neoStore.getLabelRepository().getHighId(), neoStore.getRelationshipTypeRepository().getHighId(), countsUpdater, AUTO));
        // We're done, do some final logging about it
        long totalTimeMillis = currentTimeMillis() - startTime;
        executionMonitor.done(totalTimeMillis, format("%n") + storeUpdateMonitor.toString() + format("%n") + "Peak memory usage: " + bytes(peakMemoryUsage));
        log.info("Import completed, took " + Format.duration(totalTimeMillis) + ". " + storeUpdateMonitor);
    } catch (Throwable t) {
        log.error("Error during import", t);
        throw Exceptions.launderedException(IOException.class, t);
    } finally {
        if (nodeRelationshipCache != null) {
            nodeRelationshipCache.close();
        }
        if (nodeLabelsCache != null) {
            nodeLabelsCache.close();
        }
    }
}
Also used : NodeLabelsCache(org.neo4j.unsafe.impl.batchimport.cache.NodeLabelsCache) IdMapper(org.neo4j.unsafe.impl.batchimport.cache.idmapping.IdMapper) InputRelationship(org.neo4j.unsafe.impl.batchimport.input.InputRelationship) CountsAccessor(org.neo4j.kernel.impl.api.CountsAccessor) PrimitiveLongIterator(org.neo4j.collection.primitive.PrimitiveLongIterator) Collector(org.neo4j.unsafe.impl.batchimport.input.Collector) InputCache(org.neo4j.unsafe.impl.batchimport.input.InputCache) NodeRelationshipCache(org.neo4j.unsafe.impl.batchimport.cache.NodeRelationshipCache) InputNode(org.neo4j.unsafe.impl.batchimport.input.InputNode) BatchingNeoStores(org.neo4j.unsafe.impl.batchimport.store.BatchingNeoStores) IdGenerator(org.neo4j.unsafe.impl.batchimport.cache.idmapping.IdGenerator) IOException(java.io.IOException) StatsProvider(org.neo4j.unsafe.impl.batchimport.stats.StatsProvider) RelationshipStore(org.neo4j.kernel.impl.store.RelationshipStore) IoMonitor(org.neo4j.unsafe.impl.batchimport.store.io.IoMonitor)

Aggregations

IOException (java.io.IOException)1 PrimitiveLongIterator (org.neo4j.collection.primitive.PrimitiveLongIterator)1 CountsAccessor (org.neo4j.kernel.impl.api.CountsAccessor)1 RelationshipStore (org.neo4j.kernel.impl.store.RelationshipStore)1 NodeLabelsCache (org.neo4j.unsafe.impl.batchimport.cache.NodeLabelsCache)1 NodeRelationshipCache (org.neo4j.unsafe.impl.batchimport.cache.NodeRelationshipCache)1 IdGenerator (org.neo4j.unsafe.impl.batchimport.cache.idmapping.IdGenerator)1 IdMapper (org.neo4j.unsafe.impl.batchimport.cache.idmapping.IdMapper)1 Collector (org.neo4j.unsafe.impl.batchimport.input.Collector)1 InputCache (org.neo4j.unsafe.impl.batchimport.input.InputCache)1 InputNode (org.neo4j.unsafe.impl.batchimport.input.InputNode)1 InputRelationship (org.neo4j.unsafe.impl.batchimport.input.InputRelationship)1 StatsProvider (org.neo4j.unsafe.impl.batchimport.stats.StatsProvider)1 BatchingNeoStores (org.neo4j.unsafe.impl.batchimport.store.BatchingNeoStores)1 IoMonitor (org.neo4j.unsafe.impl.batchimport.store.io.IoMonitor)1