Radoslaw Busz created SPARK-37147:
-------------------------------------
Summary: MetricsReporter producing NullPointerException when
element 'triggerExecution' not present in Map[]
Key: SPARK-37147
URL: https://issues.apache.org/jira/browse/SPARK-37147
Project: Spark
Issue Type: Bug
Components: Structured Streaming
Affects Versions: 3.2.0
Reporter: Radoslaw Busz
The exception occurs in MetricsReporter when it tries to register gauges using
lastProgress of each stream. This problem was partially fixed in
https://issues.apache.org/jira/browse/SPARK-22975 but in introduced a NPE when
'triggerExecution' element is not present in durationMS map:
{code:java}
registerGauge("latency", _.durationMs.get("triggerExecution").longValue(), 0L)
{code}
I find it difficult to reproduce this every time but it happens every few
restarts when structured streaming uses a slow event source (very little or no
events). In my case it breaks metric reporting via Codehale/Dropwizard and
generates multiple stacktraces such as:
{code:java}
21/09/16 09:51:36 ERROR ScheduledReporter: Exception thrown from
GangliaReporter#report. Exception was suppressed.
java.lang.NullPointerException
at
org.apache.spark.sql.execution.streaming.MetricsReporter.$anonfun$new$3(MetricsReporter.scala:43)
at
org.apache.spark.sql.execution.streaming.MetricsReporter.$anonfun$new$3$adapted(MetricsReporter.scala:43)
at scala.Option.map(Option.scala:230)
at
org.apache.spark.sql.execution.streaming.MetricsReporter$$anon$1.getValue(MetricsReporter.scala:68)
at
com.codahale.metrics.ganglia.GangliaReporter.reportGauge(GangliaReporter.java:353)
at
com.codahale.metrics.ganglia.GangliaReporter.report(GangliaReporter.java:240)
at
com.codahale.metrics.ScheduledReporter.report(ScheduledReporter.java:237)
at
com.codahale.metrics.ScheduledReporter.lambda$start$0(ScheduledReporter.java:177)
at
java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308)
at
java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)
at
java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294)
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
{code}
I'm happy to implement a fix.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
