Come inviare le metriche Spark a Graphite su cluster autonomo? - scintilla di apache, metrica, grafite

Provo a inviare le metriche Spark a Graphite usando la seguente configurazione:

*.sink.graphite.class=org.apache.spark.metrics.sink.GraphiteSink
*.sink.graphite.host=85.10.206.170
*.sink.graphite.port=2003
*.sink.graphite.period=1
*.sink.graphite.unit=minutes

# Enable jvm source for instance master, worker, driver and executor
master.source.jvm.class=org.apache.spark.metrics.source.JvmSource

worker.source.jvm.class=org.apache.spark.metrics.source.JvmSource

driver.source.jvm.class=org.apache.spark.metrics.source.JvmSource

executor.source.jvm.class=org.apache.spark.metrics.source.JvmSource

application.source.jvm.class=org.apache.spark.metrics.source.JvmSource

salvato su /data/configurations/metrics.properties.

Inoltro la mia domanda con queste proprietà:

--files=/data/configuration/metrics.properties --conf spark.metrics.conf=metrics.properties

Ottengo il seguente errore:

com.test.MyApp: metrics.properties (No such file or directory)
java.io.FileNotFoundException: metrics.properties (No such file or directory)
at java.io.FileInputStream.open0(Native Method) ~[?:1.8.0_45]
at java.io.FileInputStream.open(FileInputStream.java:195) ~[?:1.8.0_45]
at java.io.FileInputStream.<init>(FileInputStream.java:138) ~[?:1.8.0_45]
at java.io.FileInputStream.<init>(FileInputStream.java:93) ~[?:1.8.0_45]
at org.apache.spark.metrics.MetricsConfig$$anonfun$1.apply(MetricsConfig.scala:50) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.metrics.MetricsConfig$$anonfun$1.apply(MetricsConfig.scala:50) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at scala.Option.map(Option.scala:145) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.metrics.MetricsConfig.initialize(MetricsConfig.scala:50) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.metrics.MetricsSystem.<init>(MetricsSystem.scala:93) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.metrics.MetricsSystem$.createMetricsSystem(MetricsSystem.scala:222) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.SparkEnv$.create(SparkEnv.scala:361) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.SparkEnv$.createDriverEnv(SparkEnv.scala:188) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.SparkContext.createSparkEnv(SparkContext.scala:267) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.SparkContext.<init>(SparkContext.scala:424) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.streaming.StreamingContext$.createNewSparkContext(StreamingContext.scala:842) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.streaming.StreamingContext.<init>(StreamingContext.scala:80) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]
at org.apache.spark.streaming.api.java.JavaStreamingContext.<init>(JavaStreamingContext.scala:133) ~[spark-assembly-1.4.1-hadoop2.4.0.jar:1.4.1]

Dove mi sbaglio?

risposte:

tl; dr spark.metrics.conf dovrebbe essere un percorso assoluto.

NB: una stella (*) si riferisce qualunque sorgente metrica disponibile in Spark che può essere driver, executor, esterno shuffleService, master, applications, worker, mesos_cluster.

SUGGERIMENTO: puoi accedere alle metriche utilizzando gli URL corrispondenti per i servizi, ad es. 4040 per il driver, 8080 per il master e le applicazioni di Spark Standalone, utilizzando http://localhost:[port]/metrics/json/ URL.