Search in sources :

Example 1 with BulkWriteRequestBuilder

use of com.mongodb.BulkWriteRequestBuilder in project mongo-hadoop by mongodb.

the class MongoOutputCommitter method commitTask.

public void commitTask(final CompatUtils.TaskAttemptContext taskContext) throws IOException {
    LOG.info("Committing task.");
    collection = MongoConfigUtil.getOutputCollection(taskContext.getConfiguration());
    // Get temporary file.
    Path tempFilePath = getTaskAttemptPath(taskContext);
    LOG.info("Committing from temporary file: " + tempFilePath.toString());
    long filePos = 0, fileLen;
    FSDataInputStream inputStream = null;
    try {
        FileSystem fs = FileSystem.get(taskContext.getConfiguration());
        inputStream = fs.open(tempFilePath);
        fileLen = fs.getFileStatus(tempFilePath).getLen();
    } catch (IOException e) {
        LOG.error("Could not open temporary file for committing", e);
        cleanupAfterCommit(inputStream, taskContext);
        throw e;
    }
    int maxDocs = MongoConfigUtil.getBatchSize(taskContext.getConfiguration());
    int curBatchSize = 0;
    BulkWriteOperation bulkOp;
    if (MongoConfigUtil.isBulkOrdered(taskContext.getConfiguration())) {
        bulkOp = collection.initializeOrderedBulkOperation();
    } else {
        bulkOp = collection.initializeUnorderedBulkOperation();
    }
    // Read Writables out of the temporary file.
    BSONWritable bw = new BSONWritable();
    MongoUpdateWritable muw = new MongoUpdateWritable();
    while (filePos < fileLen) {
        try {
            // Determine writable type, and perform corresponding operation
            // on MongoDB.
            int mwType = inputStream.readInt();
            if (MongoWritableTypes.BSON_WRITABLE == mwType) {
                bw.readFields(inputStream);
                bulkOp.insert(new BasicDBObject(bw.getDoc().toMap()));
            } else if (MongoWritableTypes.MONGO_UPDATE_WRITABLE == mwType) {
                muw.readFields(inputStream);
                DBObject query = new BasicDBObject(muw.getQuery().toMap());
                DBObject modifiers = new BasicDBObject(muw.getModifiers().toMap());
                BulkWriteRequestBuilder writeBuilder = bulkOp.find(query);
                if (muw.isReplace()) {
                    writeBuilder.replaceOne(modifiers);
                } else if (muw.isUpsert()) {
                    BulkUpdateRequestBuilder updateBuilder = writeBuilder.upsert();
                    if (muw.isMultiUpdate()) {
                        updateBuilder.update(modifiers);
                    } else {
                        updateBuilder.updateOne(modifiers);
                    }
                } else {
                    // No-upsert update.
                    if (muw.isMultiUpdate()) {
                        writeBuilder.update(modifiers);
                    } else {
                        writeBuilder.updateOne(modifiers);
                    }
                }
            } else {
                throw new IOException("Unrecognized type: " + mwType);
            }
            filePos = inputStream.getPos();
            // operation to be performed for the Task.
            if (++curBatchSize >= maxDocs || filePos >= fileLen) {
                try {
                    bulkOp.execute();
                } catch (MongoException e) {
                    LOG.error("Could not write to MongoDB", e);
                    throw e;
                }
                bulkOp = collection.initializeOrderedBulkOperation();
                curBatchSize = 0;
                // Signal progress back to Hadoop framework so that we
                // don't time out.
                taskContext.progress();
            }
        } catch (IOException e) {
            LOG.error("Error reading from temporary file", e);
            throw e;
        }
    }
    cleanupAfterCommit(inputStream, taskContext);
}
Also used : Path(org.apache.hadoop.fs.Path) BulkWriteOperation(com.mongodb.BulkWriteOperation) MongoException(com.mongodb.MongoException) IOException(java.io.IOException) BasicDBObject(com.mongodb.BasicDBObject) DBObject(com.mongodb.DBObject) BSONWritable(com.mongodb.hadoop.io.BSONWritable) BasicDBObject(com.mongodb.BasicDBObject) BulkWriteRequestBuilder(com.mongodb.BulkWriteRequestBuilder) FileSystem(org.apache.hadoop.fs.FileSystem) MongoUpdateWritable(com.mongodb.hadoop.io.MongoUpdateWritable) FSDataInputStream(org.apache.hadoop.fs.FSDataInputStream) BulkUpdateRequestBuilder(com.mongodb.BulkUpdateRequestBuilder)

Aggregations

BasicDBObject (com.mongodb.BasicDBObject)1 BulkUpdateRequestBuilder (com.mongodb.BulkUpdateRequestBuilder)1 BulkWriteOperation (com.mongodb.BulkWriteOperation)1 BulkWriteRequestBuilder (com.mongodb.BulkWriteRequestBuilder)1 DBObject (com.mongodb.DBObject)1 MongoException (com.mongodb.MongoException)1 BSONWritable (com.mongodb.hadoop.io.BSONWritable)1 MongoUpdateWritable (com.mongodb.hadoop.io.MongoUpdateWritable)1 IOException (java.io.IOException)1 FSDataInputStream (org.apache.hadoop.fs.FSDataInputStream)1 FileSystem (org.apache.hadoop.fs.FileSystem)1 Path (org.apache.hadoop.fs.Path)1