You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@yunikorn.apache.org by "Weiwei Yang (Jira)" <ji...@apache.org> on 2021/07/26 21:14:00 UTC

[jira] [Resolved] (YUNIKORN-757) Discover daemonset pods and tag with ignoreUnschedulableNodes attributes

     [ https://issues.apache.org/jira/browse/YUNIKORN-757?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Weiwei Yang resolved YUNIKORN-757.
----------------------------------
    Fix Version/s: 1.0.0
       Resolution: Fixed

> Discover daemonset pods and tag with ignoreUnschedulableNodes attributes
> ------------------------------------------------------------------------
>
>                 Key: YUNIKORN-757
>                 URL: https://issues.apache.org/jira/browse/YUNIKORN-757
>             Project: Apache YuniKorn
>          Issue Type: Sub-task
>            Reporter: Weiwei Yang
>            Assignee: Ting Yao,Huang
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 1.0.0
>
>
> Define a tag for allocationAsk
> {code}
> message AllocationAsk {
>  // Allocation key is used by both of scheduler and RM to track allocations.
>  // It doesn't have to be same as RM's internal allocation id (such as Pod name of K8s or ContainerID of YARN).
>  // Allocations from the same AllocationAsk which are returned to the RM at the same time will have the same allocationKey.
>  // The request is considered an update of the existing AllocationAsk if an ALlocationAsk with the same allocationKey 
>  // already exists.
>  string allocationKey = 1;
>  // The application ID this allocation ask belongs to
>  string applicationID = 2;
>  // The partition the application belongs to
>  string partitionName = 3;
>  // The amount of resources per ask
>  Resource resourceAsk = 4;
>  // Maximum number of allocations
>  int32 maxAllocations = 5;
>  // Priority of ask
>  Priority priority = 6;
>  // Execution timeout: How long this allocation will be terminated (by scheduler)
>  // once allocated by scheduler, 0 or negative value means never expire.
>  int64 executionTimeoutMilliSeconds = 7;
>  // A set of tags for this spscific AllocationAsk. Allocation level tags are used in placing this specific
>  // ask on nodes in the cluster. These tags are used in the PlacementConstraints.
>  // These tags are optional.
>  map<string, string> tags = 8;
>  // The name of the TaskGroup this ask belongs to
>  string taskGroupName = 9;
>  // Is this a placeholder ask (true) or a real ask (false), defaults to false
>  // ignored if the taskGroupName is not set
>  bool placeholder = 10;
> }
> {code}
> yunikorn.apache.org/ignoreUnschedulableNodes: true
> default value is false.
> For every daemonset pod, set this tag with the value true. So the core side can pick up to ignore the unschedulable nodes and continue to schedule onto them.
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@yunikorn.apache.org
For additional commands, e-mail: issues-help@yunikorn.apache.org