You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@ignite.apache.org by "Dmitriy Setrakyan (JIRA)" <ji...@apache.org> on 2018/05/30 21:49:00 UTC

[jira] [Commented] (IGNITE-8649) Service assignments should be recalculated when deployment fails

    [ https://issues.apache.org/jira/browse/IGNITE-8649?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16495732#comment-16495732 ] 

Dmitriy Setrakyan commented on IGNITE-8649:
-------------------------------------------

This is a very common scenario. Some nodes purposely may not even have the service in the class path, in which case the decerialization of the service will fail. In this case, such nodes should be excluded from the service topology and the service should be deployed on another node.

Currently, I believe that Ignite nodes will try to deserialize a service during the discovery process and fail, even if they do not participate in service deployment at all. This behavior is wrong and should be fixed.

> Service assignments should be recalculated when deployment fails
> ----------------------------------------------------------------
>
>                 Key: IGNITE-8649
>                 URL: https://issues.apache.org/jira/browse/IGNITE-8649
>             Project: Ignite
>          Issue Type: Improvement
>          Components: managed services
>            Reporter: Denis Mekhanikov
>            Priority: Major
>              Labels: iep-17
>
> When a node fails to instantiate and initialize a service during deployment, coordinator should try to assign this service to another node.
> Deployment should be considered failed only when all suitable nodes failed to deploy the service.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)