You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Benjamin Teke (Jira)" <ji...@apache.org> on 2021/05/11 15:07:00 UTC
[jira] [Updated] (YARN-10505) Extend the maximum-capacity property
to react to weight mode changes
[ https://issues.apache.org/jira/browse/YARN-10505?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Benjamin Teke updated YARN-10505:
---------------------------------
Description:
The property root.users.maximum-capacity could mean the following things:
* Relative Percentage: maximum capacity relative to its parent. If it’s set to 50, then it means that the capacity is capped with respect to the parent.
* Absolute Percentage: maximum capacity expressed as a percentage of the overall cluster capacity.
* Percentages of different resource types: this would refer to vCores, memory, GPU, etc... Similarly to the single percentage value, this could either mean percentage of the parent or percentage of the overall cluster resource.
* Absolute limit: explicit definition of vCores and memory like vcores=20, memory-mb=16384. This case will be covered in YARN-9936.
Note that Fair Scheduler supports the following settings:
* Single percentage (absolute)
* Two percentages (absolute)
* Absolute resources
It is recommended that all three formats are supported for maximum-capacity after introducing weight mode.
was:
The property root.users.maximum-capacity could mean the following things:
* Relative Percentage: maximum capacity relative to its parent. If it’s set to 50, then it means that the capacity is capped with respect to the parent.
* Absolute Percentage: maximum capacity expressed as a percentage of the overall cluster capacity.
* Percentages of different resource types: this would refer to vCores, memory, GPU, etc... Similarly to the single percentage value, this could either mean percentage of the parent or percentage of the overall cluster resource.
* Absolute limit: explicit definition of vCores and memory like vcores=20, memory-mb=16384.
Note that Fair Scheduler supports the following settings:
* Single percentage (absolute)
* Two percentages (absolute)
* Absolute resources
It is recommended that all three formats are supported for maximum-capacity after introducing weight mode. The final form of the configuration for example could look like this:
root.users.maximum-capacity = 100% - single percentage
root.users.maximum-capacity = (vcores=100%, memory-mb=100%) - two percentages
root.users.maximum-capacity = (vcores=10, memory-mb=10000mb) - absolute
> Extend the maximum-capacity property to react to weight mode changes
> --------------------------------------------------------------------
>
> Key: YARN-10505
> URL: https://issues.apache.org/jira/browse/YARN-10505
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Benjamin Teke
> Priority: Major
>
> The property root.users.maximum-capacity could mean the following things:
> * Relative Percentage: maximum capacity relative to its parent. If it’s set to 50, then it means that the capacity is capped with respect to the parent.
> * Absolute Percentage: maximum capacity expressed as a percentage of the overall cluster capacity.
> * Percentages of different resource types: this would refer to vCores, memory, GPU, etc... Similarly to the single percentage value, this could either mean percentage of the parent or percentage of the overall cluster resource.
> * Absolute limit: explicit definition of vCores and memory like vcores=20, memory-mb=16384. This case will be covered in YARN-9936.
>
> Note that Fair Scheduler supports the following settings:
> * Single percentage (absolute)
> * Two percentages (absolute)
> * Absolute resources
>
> It is recommended that all three formats are supported for maximum-capacity after introducing weight mode.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org