Doubt on Scheduling Logs

[Prabhu Joseph <pr...@gmail.com>]

Hi All,

   While debugging an YARN issue on one of our customer sites
(hadoop-3.2.1), have seen below ResourceManager logs which look buggy -
The root queue level used resource is zero whereas the node level the usage
is more than 50GB.  Suspecting this is causing our AutoScaler to scale down
the worker nodes thinking the cluster is idle.

Is there any bug in this area of code, where the root queue resource is
shown as zero even though the cluster is occupied.

Thanks,
Prabhu Joseph


2022-07-02 19:00:01,831 INFO
org.apache.hadoop.yarn.server.resourcemanager.rmcontainer.RMContainerImpl
(SchedulerEventDispatcher:Event Processor):
container_e122_1656742185869_0034_01_000054 Container Transitioned from NEW
to RESERVED

2022-07-02 19:00:01,831 INFO
org.apache.hadoop.yarn.server.resourcemanager.scheduler.common.fica.FiCaSchedulerApp
(SchedulerEventDispatcher:Event Processor): Reserved
container=container_e122_1656742185869_0034_01_000054, on node=host:
host1:8041 #containers=14 available=<memory:4096, vCores:-6, max
vCores:0> *used=<memory:53248,
vCores:14>* with resource=<memory:5120, vCores:1>

2022-07-02 19:00:01,831 INFO
org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.ParentQueue
(SchedulerEventDispatcher:Event Processor): assignedContainer *queue=root
usedCapacity=0.0 absoluteUsedCapacity=0.0 used=<memory:0, vCores:0>
cluster=<memory:458752, vCores:64>*