[
https://issues.apache.org/jira/browse/YARN-10565?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Benjamin Teke updated YARN-10565:
---------------------------------
Description:
In YARN-10504 weight mode support was introduced to CS. This jira is a followup
to simplify and restructure the initialization, so that the weight
calculation/absolute/percentage mode is easier to understand and modify.
To be refactored:
* In ParentQueue.java#1099 the error message should be more specific, instead
of the {{LOG.error("Fatal issue found: e", e);}}
*- AutoCreatedLeafQueue.clearConfigurableFields should clear NORMALIZED_WEIGHT
just to be on the safe side-
*- Uncomment the commented assertions in
TestCapacitySchedulerAutoCreatedQueueBase.validateEffectiveMinResource-
*- Check whether the assertion modification in TestRMWebServices is absolutely
necessary or could be hiding a bug.-
* -Same for TestRMWebServicesForCSWithPartitions.java-
Additional information:
The original flow was modified to allow the dynamic weight-capacity
calculation.
This resulted in a new flow, which is now harder to understand.
With a cleanup it could be made simpler, the duplicate calculations could be
avoided.
The changed functionality should either be explained (if deemed correct) or
fixed (see YARN-10590).
Investigate how the CS reinit works, it could contain some possibly redundant
initialization code fragments.
Note: Since most of the items were completed in other refactor items, only the
first one is being patched here.
was:
In YARN-10504 weight mode support was introduced to CS. This jira is a followup
to simplify and restructure the initialization, so that the weight
calculation/absolute/percentage mode is easier to understand and modify.
To be refactored:
* In ParentQueue.java#1099 the error message should be more specific, instead
of the {{LOG.error("Fatal issue found: e", e);}}
* AutoCreatedLeafQueue.clearConfigurableFields should clear NORMALIZED_WEIGHT
just to be on the safe side
* Uncomment the commented assertions in
TestCapacitySchedulerAutoCreatedQueueBase.validateEffectiveMinResource
* Check whether the assertion modification in TestRMWebServices is absolutely
necessary or could be hiding a bug.
* Same for TestRMWebServicesForCSWithPartitions.java
Additional information:
The original flow was modified to allow the dynamic weight-capacity
calculation.
This resulted in a new flow, which is now harder to understand.
With a cleanup it could be made simpler, the duplicate calculations could be
avoided.
The changed functionality should either be explained (if deemed correct) or
fixed (see YARN-10590).
Investigate how the CS reinit works, it could contain some possibly redundant
initialization code fragments.
> Refactor CS queue initialization to simplify weight mode calculation
> --------------------------------------------------------------------
>
> Key: YARN-10565
> URL: https://issues.apache.org/jira/browse/YARN-10565
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Benjamin Teke
> Assignee: Benjamin Teke
> Priority: Major
> Labels: pull-request-available
> Time Spent: 50m
> Remaining Estimate: 0h
>
> In YARN-10504 weight mode support was introduced to CS. This jira is a
> followup to simplify and restructure the initialization, so that the weight
> calculation/absolute/percentage mode is easier to understand and modify.
> To be refactored:
> * In ParentQueue.java#1099 the error message should be more specific, instead
> of the {{LOG.error("Fatal issue found: e", e);}}
> *- AutoCreatedLeafQueue.clearConfigurableFields should clear
> NORMALIZED_WEIGHT just to be on the safe side-
> *- Uncomment the commented assertions in
> TestCapacitySchedulerAutoCreatedQueueBase.validateEffectiveMinResource-
> *- Check whether the assertion modification in TestRMWebServices is
> absolutely necessary or could be hiding a bug.-
> * -Same for TestRMWebServicesForCSWithPartitions.java-
> Additional information:
> The original flow was modified to allow the dynamic weight-capacity
> calculation.
> This resulted in a new flow, which is now harder to understand.
> With a cleanup it could be made simpler, the duplicate calculations could be
> avoided.
> The changed functionality should either be explained (if deemed correct) or
> fixed (see YARN-10590).
> Investigate how the CS reinit works, it could contain some possibly redundant
> initialization code fragments.
> Note: Since most of the items were completed in other refactor items, only
> the first one is being patched here.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
