[ 
https://issues.apache.org/jira/browse/FLINK-24947?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17488631#comment-17488631
 ] 

Yuan Huang  commented on FLINK-24947:
-------------------------------------

Thank you for your reply. Yes, I think so. And I believe even though Flink uses 
the head service, the TaskManager also can not be reused if Task managers do 
not want to query the service for the "jobmanager-rpc" port because the port 
number is not fixed.

I can imagine two reasons why the TaskManager does not want to access the 
service:
 # Involving Kube-client in all TaskManagers will put too much pressure on the 
Kubernetes API server.
 # In non-HA mode, the StandaloneHaService saves a fixed job manager address 
(IP and port) at the beginning. It is not natural to change the address when 
the TaskManagers lose connection to the JobManager. 

Am I understanding correctly?

> Support host network for native K8s integration
> -----------------------------------------------
>
>                 Key: FLINK-24947
>                 URL: https://issues.apache.org/jira/browse/FLINK-24947
>             Project: Flink
>          Issue Type: New Feature
>          Components: Deployment / Kubernetes
>            Reporter: liuzhuo
>            Assignee: liuzhuo
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 1.15.0
>
>
> For the use of flink on k8s, for performance considerations, it is important 
> to choose a CNI plug-in. Usually we have two environments: Managed and 
> UnManaged.
>   Managed: Cloud vendors usually provide very efficient CNI plug-ins, we 
> don’t need to care about network performance issues
>   UnManaged: On self-built K8s clusters, CNI plug-ins are usually optional, 
> similar to Flannel and Calcico, but such software network cards usually lose 
> some performance or require some additional network strategies.
> For an unmanaged environment, if we also want to achieve the best network 
> performance, should we support the *HostNetWork* model?
> Use the host network to achieve the best performance



--
This message was sent by Atlassian Jira
(v8.20.1#820001)

Reply via email to