关于java.lang.NoClassDefFoundError：无法初始化类org.xerial.snappy.Snappy

我尝试编译，测试和运行junit测试时出现错误。

我想使用DataFrames加载本地Avro文件，但出现异常：

org.xerial.snappy.SnappyError: [FAILED_TO_LOAD_NATIVE_LIBRARY] null

我根本没有使用Cassandra，涉及的jar版本是：

<properties>
    <!-- Generic properties -->
    <java.version>1.7</java.version>
    <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
    <project.reporting.outputEncoding>UTF-8</project.reporting.outputEncoding>
    <!--  Dependency versions -->
    <maven.compiler.source>1.7</maven.compiler.source>
    <maven.compiler.target>1.7</maven.compiler.target>
    <scala.version>2.10.4</scala.version>
    <junit.version>4.11</junit.version>
    <slf4j.version>1.7.12</slf4j.version>
    <spark.version>1.5.0-cdh5.5.2</spark.version>
    <databricks.version>1.5.0</databricks.version>
    <json4s-native.version>3.5.0</json4s-native.version>
    <spark-avro.version>2.0.1</spark-avro.version>
</properties>

这些是依赖项：

<dependencies>

    <dependency>
        <groupId>org.json4s</groupId>
        <artifactId>json4s-native_2.10</artifactId>
        <version>${json4s-native.version}</version>
    </dependency>
    <dependency>
        <groupId>junit</groupId>
        <artifactId>junit</artifactId>
        <version>${junit.version}</version>
        <scope>test</scope>
    </dependency>
    <dependency>
        <groupId>org.apache.spark</groupId>
        <artifactId>spark-sql_2.10</artifactId>
        <version>${spark.version}</version>
    </dependency>
    <dependency>
        <groupId>com.databricks</groupId>
        <artifactId>spark-csv_2.10</artifactId>
        <version>${databricks.version}</version>
        <exclusions>
            <exclusion>
                <groupId>org.xerial.snappy</groupId>
                <artifactId>snappy-java</artifactId>
            </exclusion>
        </exclusions>
    </dependency>

    <dependency>
        <groupId>org.xerial.snappy</groupId>
        <artifactId>snappy-java</artifactId>
        <version>1.0.4.1</version>
        <scope>compile</scope>
    </dependency>
    <dependency>
        <groupId>com.databricks</groupId>
        <artifactId>spark-avro_2.10</artifactId>
        <version>${spark-avro.version}</version>
    </dependency>
    <!-- https://mvnrepository.com/artifact/log4j/log4j -->
    <dependency>
        <groupId>log4j</groupId>
        <artifactId>log4j</artifactId>
        <version>1.2.17</version>
    </dependency>
</dependencies>

我试图用编译该项目

mvn clean install -Dorg.xerial.snappy.lib.name=libsnappyjava.jnlib -Dorg.xerial.snappy.tempdir=/tmp

在/ tmp内复制jar之前，没有运气。

$ ls -lt /tmp/
total 1944
...27 dic 13:01 snappy-java-1.0.4.jar

这是代码：

import org.apache.spark.rdd.RDD
import org.apache.spark.sql.{DataFrame, Row, SQLContext, SaveMode}
import org.apache.spark.{SparkConf, SparkContext}
import com.databricks.spark.avro._
import java.io._

//auxiliary function
def readRawData(pathToResources: String, sqlContext: SQLContext, rawFormat: String = "json"): DataFrame = {
val a: DataFrame = rawFormat match {
  case "avro" => sqlContext.read.avro(pathToResources)
  case "json" => sqlContext.read.json(pathToResources)
  case _ => throw new Exception("Format not supported, use AVRO or JSON instead.")
}
val b: DataFrame = a.filter("extraData.type = 'data'")
val c: DataFrame = a.select("extraData.topic", "extraData.timestamp",
  "extraData.sha1Hex", "extraData.filePath", "extraData.fileName",
  "extraData.lineNumber", "extraData.type",
  "message")

val indexForMessage: Int = c.schema.fieldIndex("message")
val result: RDD[Row] = c.rdd.filter(r =>
  !r.anyNull match {
    case true => true
    case false => false
  }
).flatMap(r => {
  val metadata: String = r.toSeq.slice(0, indexForMessage).mkString(",")
  val lines = r.getString(indexForMessage).split("\n")
  lines.map(l => Row.fromSeq(metadata.split(",").toSeq ++ Seq(l)))
})
sqlContext.createDataFrame(result, c.schema)
}//readRawData


def validate(rawFlumeData : String = "FlumeData.1482407196579",fileNamesToBeDigested : String = "fileNames-to-be-digested.txt", sqlContext: SQLContext,sc:SparkContext) : Boolean = {

val result : Boolean = true

  sqlContext.sparkContext.hadoopConfiguration.set("avro.mapred.ignore.inputs.without.extension", "false")

val rawDF : DataFrame = readRawData(rawFlumeData, sqlContext, rawFormat = "avro")

rawDF.registerTempTable("RAW")
//this line provokes the exception! cannot load snappy jar file!
val arrayRows : Array[org.apache.spark.sql.Row] = sqlContext.sql("SELECT distinct fileName as filenames FROM RAW GROUP BY fileName").collect()

val arrayFileNames : Array[String] = arrayRows.map(row=>row.getString(0))

val fileNamesDigested = "fileNames-AVRO-1482407196579.txt"

val pw = new PrintWriter(new File(fileNamesDigested))

for (filename <-arrayFileNames) pw.write(filename + "\n")

pw.close

val searchListToBeDigested : org.apache.spark.rdd.RDD[String] = sc.textFile(fileNamesToBeDigested)

//creo un map con valores como éstos: Map(EUR_BACK_SWVOL_SMILE_GBP_20160930.csv -> 0, UK_SC_equities_20160930.csv -> 14,...
//val mapFileNamesToBeDigested: Map[String, Long] = searchListToBeDigested.zipWithUniqueId().collect().toMap

val searchFilesAVRODigested = sc.textFile(fileNamesDigested)

val mapFileNamesAVRODigested: Map[String, Long] = searchFilesAVRODigested.zipWithUniqueId().collect().toMap

val pwResults = new PrintWriter(new File("validation-results.txt"))

//Hay que guardar el resultado en un fichero de texto, en algún lado...
val buffer = StringBuilder.newBuilder

//Me traigo los resultados al Driver.
val listFilesToBeDigested = searchListToBeDigested.map {line =>
  val resultTemp = mapFileNamesAVRODigested.getOrElse(line,"NOT INGESTED!")
  var resul = ""
  if (resultTemp == "NOT INGESTED!"){
    resul = "File " + line + " " + resultTemp + "\n"
  }
  else{
    resul = "File " + line + " " + " is INGESTED!" + "\n"
  }
  resul
}.collect()

//añado los datos al buffer
listFilesToBeDigested.foreach(buffer.append(_))
//guardo el contenido del buffer en el fichero de texto de salida.
pwResults.write(buffer.toString)
pwResults.close
//this boolean must return false in case of a exception or error...
result
}//

这是单元测试代码：

private[validation] class ValidateInputCSVFilesTest {

//AS YOU CAN SEE, I do not WANT to use snappy at all!
val conf = new SparkConf()
.setAppName("ValidateInputCSVFilesTest")
.setMaster("local[2]")
.set("spark.driver.allowMultipleContexts", "true")
.set("spark.driver.host", "127.0.0.1")
.set("spark.io.compression.codec", "lzf")

val sc = new SparkContext(conf)
val sqlContext = new org.apache.spark.sql.SQLContext(sc)

val properties : Properties = new Properties()
 properties.setProperty("frtb.input.csv.validation.avro","./src/test/resources/avro/FlumeData.1482407196579")
properties.setProperty("frtb.input.csv.validation.list.files","./src/test/resources/fileNames-to-be-digested.txt")
 import sqlContext.implicits._

sqlContext.sparkContext.hadoopConfiguration.set("avro.mapred.ignore.inputs.without.extension", "false")

@Test
def testValidateInputFiles() = {

//def validate(rawFlumeData : String = "FlumeData.1482407196579",fileNamesToBeDigested : String = "fileNames-to-be-digested.txt", sqlContext: SQLContext)
val rawFlumeData = properties.getProperty("frtb.input.csv.validation.avro")
val fileNamesToBeDigested = properties.getProperty("frtb.input.csv.validation.list.files")
println("rawFlumeData  is " + rawFlumeData )
println("fileNamesToBeDigested is " + fileNamesToBeDigested )
val result : Boolean = ValidateInputCSVFiles.validate(rawFlumeData ,fileNamesToBeDigested ,sqlContext,sc)

Assert.assertTrue("Must be true...",result)

}//end of test method

}//end of unit class

我可以使用以下命令在本地spark-shell中运行完全相同的代码：

$ bin/spark-shell --packages org.json4s:json4s-native_2.10:3.5.0 --packages com.databricks:spark-csv_2.10:1.5.0 --packages com.databricks:spark-avro_2.10:2.0.1 

我还可以做些什么？

提前致谢。