HeartSaVioR commented on a change in pull request #27002: [SPARK-30346]Improve
logging when events dropped
URL: https://github.com/apache/spark/pull/27002#discussion_r361572976
##########
File path: core/src/main/scala/org/apache/spark/scheduler/AsyncEventQueue.scala
##########
@@ -167,20 +170,29 @@ private class AsyncEventQueue(
}
logTrace(s"Dropping event $event")
- val droppedCount = droppedEventsCounter.get
+ val droppedCount = droppedEventsCounter.get - lastDroppedEventsCounter
+ val lastReportTime = lastReportTimestamp.get
+ val curTime = System.currentTimeMillis()
if (droppedCount > 0) {
// Don't log too frequently
- if (System.currentTimeMillis() - lastReportTimestamp >= 60 * 1000) {
- // There may be multiple threads trying to decrease
droppedEventsCounter.
- // Use "compareAndSet" to make sure only one thread can win.
- // And if another thread is increasing droppedEventsCounter,
"compareAndSet" will fail and
- // then that thread will update it.
- if (droppedEventsCounter.compareAndSet(droppedCount, 0)) {
- val prevLastReportTimestamp = lastReportTimestamp
- lastReportTimestamp = System.currentTimeMillis()
+ if (curTime - lastReportTime >= LOGGING_INTERVAL) {
+ // There may be multiple threads trying to logging dropped events,
+ // Use 'compareAndSet' to make sure only one thread can win.
+ // After set the 'lastReportTimestamp', the next time we come here will
+ // be 60s later.
+ if (lastReportTimestamp.compareAndSet(lastReportTime, curTime)) {
+ val prevLastReportTimestamp = lastReportTimestamp.get
val previous = new java.util.Date(prevLastReportTimestamp)
+ lastDroppedEventsCounter = droppedCount
logWarning(s"Dropped $droppedCount events from $name since " +
s"${if (prevLastReportTimestamp == 0) "the application started"
else s"$previous"}.")
+ // Logging thread dump when events from appStatus was dropped
Review comment:
I'm not sure how much this helps; as current event being processed may not
be culprit to lag the overall executions.
Spark will introduce a new configuration to log the event and listener name
if the execution takes more than configured threshold; it would be more
helpful. The commit is
https://github.com/apache/spark/commit/0346afa8fc348aa1b3f5110df747a64e3b2da388
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
