[
https://issues.apache.org/jira/browse/YARN-7892?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16357876#comment-16357876
]
Weiwei Yang commented on YARN-7892:
-----------------------------------
Hi [~Naganarasimha], [~sunilg]
I think we should check both prefix/key/name/value in equals (hashcode),
otherwise how we can tell if two attributes are same? For example if we have
following attributes on h1 and h2,
{code:java}
h1: A=A1
h2: A=A2
{code}
In attribute-manager, we use node-attribute as a key that maps to a list of
nodes have this attribute,
attribute -> nodes
with this patch, the mapping will be like
{code:java}
A: [h1, h2]
{code}
but when we do placement decision, we want to put a container on only nodes
where {{A=A1}}, then how can this be done?
Regarding to [~Naganarasimha]'s concern
{quote}Else in every place we need to separately start storing with Name and
prefix what are the attributes for example 7856 is newly introducing a map to
just compare whether any duplicates exists.
{quote}
This can be resolved by an utility class method, e.g
{{NodeAttributeUtils#sameAttributeKey(NodeAttribute na1, NodeAttribute na2)}},
or something similar, not a big problem.
Does this make sense?
Thanks.
> NodeAttributePBImpl does not implement hashcode and Equals properly
> -------------------------------------------------------------------
>
> Key: YARN-7892
> URL: https://issues.apache.org/jira/browse/YARN-7892
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: resourcemanager
> Reporter: Naganarasimha G R
> Assignee: Naganarasimha G R
> Priority: Major
> Attachments: YARN-7892-YARN-3409.001.patch,
> YARN-7892-YARN-3409.002.patch
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
