[
https://issues.apache.org/jira/browse/FLINK-18067?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jiayi Liao updated FLINK-18067:
-------------------------------
Description:
Currently Flink sets {{yarnMinAllocationMB}} default value to 0, which will
crash the job in normalizing the allocation memory. There should be two minor
changes after discussion with [~fly_in_gis] in
https://github.com/apache/flink/pull/11176#discussion_r433629963:
* Make the default value to
YarnConfiguration.DEFAULT_RM_SCHEDULER_MINIMUM_ALLOCATION_MB, not 0
* Add pre-check for yarnMinAllocationMB, it should be greater than 0.
was:
Currently Flink sets {{yarnMinAllocationMB}} default value to 0, which will
crash the job in normalizing the allocation memory. There should be two minor
changes after discussion with [~fly_in_gis]:
* Make the default value to
YarnConfiguration.DEFAULT_RM_SCHEDULER_MINIMUM_ALLOCATION_MB, not 0
* Add pre-check for yarnMinAllocationMB, it should be greater than 0.
> Invalid default value for yarnMinAllocationMB in YarnClusterDescriptor
> ----------------------------------------------------------------------
>
> Key: FLINK-18067
> URL: https://issues.apache.org/jira/browse/FLINK-18067
> Project: Flink
> Issue Type: Improvement
> Components: Deployment / YARN
> Affects Versions: 1.10.1
> Reporter: Jiayi Liao
> Priority: Minor
> Labels: pull-request-available
>
> Currently Flink sets {{yarnMinAllocationMB}} default value to 0, which will
> crash the job in normalizing the allocation memory. There should be two minor
> changes after discussion with [~fly_in_gis] in
> https://github.com/apache/flink/pull/11176#discussion_r433629963:
> * Make the default value to
> YarnConfiguration.DEFAULT_RM_SCHEDULER_MINIMUM_ALLOCATION_MB, not 0
> * Add pre-check for yarnMinAllocationMB, it should be greater than 0.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
