[
https://issues.apache.org/jira/browse/YARN-3409?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15753839#comment-15753839
]
Sunil G commented on YARN-3409:
-------------------------------
Short summary from the offline discussion with [[email protected]]
[~leftnoteasy] [~kkaranasos] [~grey] [~varun_saxena]. I am taking the liberty
of summarizing. Folks, please pitch in and add more points if some points are
missed. Thank you.
*Basic requirements*
{{Attribute}} will be an entity which can be associated with a node. Few notes
about attribute.
- It will be a subclass of Node Labels
- Attribute could be considered as (key, value, type)
- Default type will be *string* for now.
- We could also have predefined types and it could be {{string}}. Later we
could add like number etc. This is to be discussed more.
- Boundary between resource and attribute should be whether its consumable or
not.
- Could use empty string to express “boolean” attributes
Create separate JIRA to define ConstraintExpression (add field to
ResourceRequest and change scheduler). Within this, we can have different
constraint types, e.g., AttributeConstraintExpression,
AffinityCostraintExpression. etc,
Few items to be discussed further more:
- Predefined attributed to created before use or not?
- Could we use existing label API / REST API or create new APIs? (An initial
consensus for this is to use existing API. We could discuss more.)
- Could have separate (dynamic) node labels for the (anti-)affinity constraints
> Add constraint node labels
> --------------------------
>
> Key: YARN-3409
> URL: https://issues.apache.org/jira/browse/YARN-3409
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: api, capacityscheduler, client
> Reporter: Wangda Tan
> Assignee: Naganarasimha G R
> Attachments: Constraint-Node-Labels-Requirements-Design-doc_v1.pdf,
> YARN-3409.WIP.001.patch
>
>
> Specify only one label for each node (IAW, partition a cluster) is a way to
> determinate how resources of a special set of nodes could be shared by a
> group of entities (like teams, departments, etc.). Partitions of a cluster
> has following characteristics:
> - Cluster divided to several disjoint sub clusters.
> - ACL/priority can apply on partition (Only market team / marke team has
> priority to use the partition).
> - Percentage of capacities can apply on partition (Market team has 40%
> minimum capacity and Dev team has 60% of minimum capacity of the partition).
> Constraints are orthogonal to partition, they’re describing attributes of
> node’s hardware/software just for affinity. Some example of constraints:
> - glibc version
> - JDK version
> - Type of CPU (x86_64/i686)
> - Type of OS (windows, linux, etc.)
> With this, application can be able to ask for resource has (glibc.version >=
> 2.20 && JDK.version >= 8u20 && x86_64).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
