Github user zsxwing commented on a diff in the pull request:
https://github.com/apache/spark/pull/4467#discussion_r29699606
--- Diff:
streaming/src/main/scala/org/apache/spark/streaming/scheduler/ReceiverTracker.scala
---
@@ -81,12 +112,14 @@ class ReceiverTracker(ssc: StreamingContext,
skipReceiverLaunch: Boolean = false
"ReceiverTracker", new ReceiverTrackerEndpoint(ssc.env.rpcEnv))
if (!skipReceiverLaunch) receiverExecutor.start()
logInfo("ReceiverTracker started")
+ trackerState = Started
}
}
/** Stop the receiver execution thread. */
def stop(graceful: Boolean): Unit = synchronized {
- if (!receiverInputStreams.isEmpty && endpoint != null) {
+ if (isTrackerStarted) {
+ trackerState = Stopping
--- End diff --
`changing trackerState to Stopping` and `calling stopReceivers()` should be
atomic, or it's still possible that `StopReceiver` won't be sent because the
`registerReceiver` method runs in another thread.
And you may need to use another lock rather than adding `synchronized` to
`registerReceiver` because it may block `ReceiverTrackerEndpoint`, which is
actually a Akka thread. If it blocks the Akka thread for a long time, some
dead-lock may happen.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
