[YARN-4134] FairScheduler preemption stops at queue level that all child queues are not over their fairshare

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Component/s: fairscheduler
• Assignee: Xianyin Xin
• Reporter: Xianyin Xin
• Created: 09/Sep/15 04:11
• Updated: 09/Mar/16 01:35
• Resolved: 04/Mar/16 18:51

Description

Now FairScheudler uses a choose-a-candidate method to select a container from leaf queues that to be preempted, in FSParentQueue.preemptContainer(),

    readLock.lock();
    try {
      for (FSQueue queue : childQueues) {
        if (candidateQueue == null ||
            comparator.compare(queue, candidateQueue) > 0) {
          candidateQueue = queue;
        }
      }
    } finally {
      readLock.unlock();
    }

    // Let the selected queue choose which of its container to preempt
    if (candidateQueue != null) {
      toBePreempted = candidateQueue.preemptContainer();
    }

a candidate child queue is selected. However, if the queue's usage isn't over it's fairshare, preemption will not happen:

    if (!preemptContainerPreCheck()) {
      return toBePreempted;
    }

A scenario:

            root
           /    \
      queue1   queue2
               /    \
          queue2.3, (  queue2.4  )

suppose there're 8 containers, and queues at any level have the same weight. queue1 takes 4 and queue2.3 takes 4, so both queue1 and queue2 are at their fairshare. Now we submit an app in queue2.4 with 4 containers needs, it should preempt 2 from queue2.3, but the candidate-containers selection procedure will stop at queue1, so none of the containers will be preempted.

Issue Links

• duplicates: YARN-3405 FairScheduler's preemption cannot happen between sibling in some case
• is depended upon by: YARN-4190 missing container information in FairScheduler preemption log.
• is related to: YARN-2154 FairScheduler: Improve preemption to preempt only those containers that would satisfy the incoming request
• is related to: YARN-4120 FSAppAttempt.getResourceUsage() should not take preemptedResource into account