Aggarwal-Raghav commented on code in PR #473:
URL: https://github.com/apache/tez/pull/473#discussion_r3037227937
##########
tez-dag/src/main/java/org/apache/tez/dag/app/TezTaskCommunicatorImpl.java:
##########
@@ -411,11 +411,12 @@ private ContainerTask getContainerTask(ContainerId
containerId) throws IOExcepti
if (getContext().isKnownContainer(containerId)) {
LOG.info("Container with id: " + containerId
+ " is valid, but no longer registered, and will be killed");
+ task = TASK_FOR_INVALID_JVM;
} else {
LOG.info("Container with id: " + containerId
- + " is invalid and will be killed");
+ + " is not yet registered, asking it to wait");
+ task = null;
Review Comment:
this is done because
`org.apache.tez.dag.app.rm.TaskSchedulerContextImpl#containerAllocated()` used
in `ZookeeperTaskScheduler.java` is just incrementing the counter, for
externally created containers it is sending `shoulddie` response but instead it
should send back null. this will help `ContainerReporter.java#callInteral()`
for loop condition to wait until ZookeeperTaskScheduler creates the container,
and registers it in the AM.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
