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

Posted to issues@ignite.apache.org by "Aleksandr (Jira)" <ji...@apache.org> on 2024/08/19 13:06:00 UTC

[jira] [Created] (IGNITE-23024) Review broadcast execution and marshallers

Aleksandr created IGNITE-23024:
----------------------------------

             Summary: Review broadcast execution and marshallers
                 Key: IGNITE-23024
                 URL: https://issues.apache.org/jira/browse/IGNITE-23024
             Project: Ignite
          Issue Type: Improvement
          Components: compute
            Reporter: Aleksandr


The problem is that when we call broadcast compute job (or any kind of job that is executed on the same node where the client code runs) in embedded move we get the following behavior:
- job argument is marshalled on client, unmarshalled on server, job called
- the result of the job is not marshalled by server when the embedded client is the same as the compute executor.

This might be ok but the behavior must be consistent: If we marshal job arguments for the same node the we should marshal job result as well. Or do not marshal at all. 

As a result of this thicket we have to define the contract and support it. As an option we can discuss "not specified".



--
This message was sent by Atlassian Jira
(v8.20.10#820010)