本文整理了Java中org.apache.spark.rdd.RDD.sparkContext
方法的一些代码示例,展示了RDD.sparkContext
的具体用法。这些代码示例主要来源于Github
/Stackoverflow
/Maven
等平台,是从一些精选项目中提取出来的代码,具有较强的参考意义,能在一定程度帮忙到你。RDD.sparkContext
方法的具体详情如下:
包路径:org.apache.spark.rdd.RDD
类名称:RDD
方法名:sparkContext
暂无
代码示例来源:origin: uber/marmaray
private void logWriteMetrics(final Optional<JavaRDD<WriteStatus>> writesStatuses) {
if (writesStatuses.isPresent() && this.dataFeedMetrics.isPresent()) {
final LongAccumulator totalCount = writesStatuses.get().rdd().sparkContext().longAccumulator();
final LongAccumulator errorCount = writesStatuses.get().rdd().sparkContext().longAccumulator();
writesStatuses.get().foreach(writeStatus -> {
errorCount.add(writeStatus.getFailedRecords().size());
totalCount.add(writeStatus.getTotalRecords());
});
this.dataFeedMetrics.get().createLongMetric(DataFeedMetricNames.ERROR_ROWCOUNT, errorCount.value(),
this.dataFeedMetricsTags);
this.dataFeedMetrics.get().createLongMetric(DataFeedMetricNames.OUTPUT_ROWCOUNT,
totalCount.value() - errorCount.value(), this.dataFeedMetricsTags);
}
}
代码示例来源:origin: com.basho.riak/spark-riak-connector-java
public void saveToRiakTS(
RiakConnector connector,
String bucketType,
String bucketName,
WriteConf writeConf,
WriteDataMapperFactory<T, RowDef> factory
) {
RiakWriter<T, RowDef> writer = RiakWriter.tsWriter(connector, bucketType, bucketName, writeConf, factory);
rdd.sparkContext().runJob(rdd, new JobFunc<>(writer), JavaApiHelper.getClassTag(Void.class));
}
代码示例来源:origin: com.basho.riak/spark-riak-connector
public void saveToRiakTS(
RiakConnector connector,
String bucketType,
String bucketName,
WriteConf writeConf,
WriteDataMapperFactory<T, RowDef> factory
) {
RiakWriter<T, RowDef> writer = RiakWriter.tsWriter(connector, bucketType, bucketName, writeConf, factory);
rdd.sparkContext().runJob(rdd, new JobFunc<>(writer), JavaApiHelper.getClassTag(Void.class));
}
代码示例来源:origin: uber/marmaray
public JavaRDD<DI> getRDD(final int filterKey) {
final long count = getCount(filterKey);
log.info("#records for :{} = {}", filterKey, count);
if (count > 0) {
return getRDD(new FilterFunction<>(filterKey));
} else {
return (new JavaSparkContext(inputRDD.rdd().sparkContext())).emptyRDD();
}
}
代码示例来源:origin: uber/marmaray
public final void execute() {
this.forkFunction.registerAccumulators(this.inputRDD.rdd().sparkContext());
// Converts JavaRDD<T> -> JavaRDD<List<Integer>, T>
JavaRDD<ForkData<DI>> forkedData = this.inputRDD.flatMap(this.forkFunction)
.persist(this.persistLevel);
final String jobName = SparkJobTracker.getJobName(this.inputRDD.rdd().sparkContext());
forkedData.setName(String.format("%s-%s", jobName, forkedData.id()));
// deliberately calling count so that DAG gets executed.
final long processedRecords = forkedData.count();
final Optional<RDDInfo> rddInfo = SparkUtil.getRddInfo(forkedData.context(), forkedData.id());
log.info("#processed records :{} name:{}", processedRecords, forkedData.name());
if (rddInfo.isPresent()) {
final long size = rddInfo.get().diskSize() + rddInfo.get().memSize();
log.info("rddInfo -> name:{} partitions:{} size:{}", forkedData.name(), rddInfo.get().numPartitions(),
size);
}
this.groupRDD = Optional.of(forkedData);
}
代码示例来源:origin: uber/marmaray
/**
* {@link #updateSinkStat(Optional)} will compute {@link SinkStat} and persist changes into {@link IMetadataManager}.
* As a part of {@link SinkStat} computation; it will compute avg record size for current run.
* @param writesStatuses
*/
private void updateSinkStat(final Optional<JavaRDD<WriteStatus>> writesStatuses) {
if (writesStatuses.isPresent()) {
final LongAccumulator avgRecordSizeCounter = writesStatuses.get().rdd().sparkContext().longAccumulator();
writesStatuses.get().foreach(
writeStatus -> {
final long writeBytes = writeStatus.getStat().getTotalWriteBytes();
final long numInserts = writeStatus.getStat().getNumWrites()
- writeStatus.getStat().getNumUpdateWrites();
if (writeBytes > 0 && numInserts > 0) {
avgRecordSizeCounter.add(writeBytes / numInserts);
}
}
);
final long avgRecordSize = (int) avgRecordSizeCounter.avg();
if (avgRecordSize > 0) {
log.info("Updating Sink Stat manager : avgRecordSize : {}", avgRecordSize);
this.sinkStatMgr.getCurrentStat().put(SinkStat.AVG_RECORD_SIZE, Long.toString(avgRecordSize));
}
}
this.sinkStatMgr.persist();
}
内容来源于网络,如有侵权,请联系作者删除!