ODIn/spark-instrumented-optimizer
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

[SPARK-33669] Wrong error message from YARN application state monitor when sc.stop in yarn client mode

Browse source 
### What changes were proposed in this pull request?
This change make InterruptedIOException to be treated as InterruptedException when closing YarnClientSchedulerBackend, which doesn't log error with "YARN application has exited unexpectedly xxx"

### Why are the changes needed?
For YarnClient mode, when stopping YarnClientSchedulerBackend, it first tries to interrupt Yarn application monitor thread. In MonitorThread.run() it catches InterruptedException to gracefully response to stopping request.

But client.monitorApplication method also throws InterruptedIOException when the hadoop rpc call is calling. In this case, MonitorThread will not know it is interrupted, a Yarn App failed is returned with "Failed to contact YARN for application xxxxx;  YARN application has exited unexpectedly with state xxxxx" is logged with error level. which confuse user a lot.

### Does this PR introduce _any_ user-facing change?
Yes

### How was this patch tested?
very simple patch, seems no need?

Closes #30617 from sqlwindspeaker/yarn-client-interrupt-monitor.

Authored-by: suqilong <suqilong@qiyi.com>
Signed-off-by: Mridul Muralidharan <mridul<at>gmail.com>
This commit is contained in:
suqilong 2020-12-09 01:21:13 -06:00 committed by Mridul Muralidharan
parent c88eddac3b
commit 48f93af9f3
2 changed files with 5 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
resource-managers/yarn/src/main/scala/org/apache/spark/deploy/yarn/Client.scala
View file

@ -1069,7 +1069,7 @@ private[spark] class Client(
logError(s"Application $appId not found.")
cleanupStagingDir()
return YarnAppReport(YarnApplicationState.KILLED, FinalApplicationStatus.KILLED, None)
case NonFatal(e) =>
case NonFatal(e) if !e.isInstanceOf[InterruptedIOException] =>
val msg = s"Failed to contact YARN for application $appId."
logError(msg, e)
// Don't necessarily clean up staging dir because status is unknown

5
resource-managers/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClientSchedulerBackend.scala
View file

@ -17,6 +17,8 @@
package org.apache.spark.scheduler.cluster
import java.io.InterruptedIOException
import scala.collection.mutable.ArrayBuffer
import org.apache.hadoop.yarn.api.records.YarnApplicationState
@ -121,7 +123,8 @@ private[spark] class YarnClientSchedulerBackend(
allowInterrupt = false
sc.stop()
} catch {
case e: InterruptedException => logInfo("Interrupting monitor thread")
case _: InterruptedException | _: InterruptedIOException =>
logInfo("Interrupting monitor thread")
}
}