[
https://issues.apache.org/jira/browse/YARN-11000?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17739679#comment-17739679
]
ASF GitHub Bot commented on YARN-11000:
---------------------------------------
tomicooler commented on code in PR #5783:
URL: https://github.com/apache/hadoop/pull/5783#discussion_r1251079277
##########
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/scheduler/capacity/AbstractParentQueue.java:
##########
@@ -1136,11 +1137,43 @@ public void refreshAfterResourceCalculation(Resource
clusterResource,
CSQueueUtils.updateConfiguredCapacityMetrics(resourceCalculator,
labelManager.getResourceByLabel(null, clusterResource),
RMNodeLabelsManager.NO_LABEL, this);
+
+ LOG.info("Refresh after resource calculation (PARENT) {}\n"
+ + "effectiveMinResource = {}\n"
+ + "effectiveMaxResource = {}\n"
+ + "capacity = {}\n"
+ + "maxCapacity = {}\n"
+ + "absoluteCapacity = {}\n"
+ + "absoluteMaxCapacity = {}",
Review Comment:
Capacity Vectors are logged from `Initializing root.Qc.Qc1`:
```
capacityVector = {=[memory-mb=50.0%,vcores=50.0%],
Lx=[memory-mb=40.0%,vcores=40.0%], Ly=[memory-mb=50.0%,vcores=50.0%]}
maxCapacityVector = {=[memory-mb=60.0%,vcores=60.0%],
Lx=[memory-mb=0.0,vcores=0.0], Ly=[memory-mb=75.0%,vcores=75.0%]}
```
and the effectiveMin/Max / label from
`calculateAndGetAMResourceLimitPerPartition`
```
2023-07-03 18:04:16,119 DEBUG [AsyncDispatcher event handler]
capacity.AbstractLeafQueue
(AbstractLeafQueue.java:calculateAndGetAMResourceLimitPerPartition(817)) -
Queue: root.Qc.Qc1, node label : , queue partition resource : <memory:0,
vCores:0>, queue current limit : <memory:0, vCores:0>, queue partition usable
resource : <memory:0, vCores:0>, amResourceLimit : <memory:0, vCores:0>
2023-07-03 18:04:16,119 DEBUG [AsyncDispatcher event handler]
capacity.AbstractLeafQueue
(AbstractLeafQueue.java:calculateAndGetAMResourceLimitPerPartition(817)) -
Queue: root.Qc.Qc1, node label : Lx, queue partition resource : <memory:573,
vCores:0>, queue current limit : <memory:0, vCores:0>, queue partition usable
resource : <memory:573, vCores:0>, amResourceLimit : <memory:1024, vCores:1>
2023-07-03 18:04:16,119 DEBUG [AsyncDispatcher event handler]
capacity.AbstractLeafQueue
(AbstractLeafQueue.java:calculateAndGetAMResourceLimitPerPartition(817)) -
Queue: root.Qc.Qc1, node label : Ly, queue partition resource : <memory:65536,
vCores:64>, queue current limit : <memory:0, vCores:0>, queue partition usable
resource : <memory:65536, vCores:64>, amResourceLimit : <memory:7168, vCores:1>
```
> Replace queue resource calculation logic in updateClusterResource
> -----------------------------------------------------------------
>
> Key: YARN-11000
> URL: https://issues.apache.org/jira/browse/YARN-11000
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: capacity scheduler
> Reporter: Andras Gyori
> Assignee: Andras Gyori
> Priority: Major
> Labels: pull-request-available
> Time Spent: 2h 10m
> Remaining Estimate: 0h
>
> YARN-10965 introduces a brand new queue calculation system. In order to
> simplify the review process, this issue replaces the current logic with the
> newly introduced one.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
