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

Posted to dev@flink.apache.org by "Gyula Fora (JIRA)" <ji...@apache.org> on 2015/10/25 11:59:27 UTC

[jira] [Created] (FLINK-2916) Add more context information for the StateBackend

Gyula Fora created FLINK-2916:
---------------------------------

             Summary: Add more context information for the StateBackend
                 Key: FLINK-2916
                 URL: https://issues.apache.org/jira/browse/FLINK-2916
             Project: Flink
          Issue Type: Improvement
          Components: Streaming
            Reporter: Gyula Fora
            Assignee: Gyula Fora
            Priority: Minor


The StateBackend currently only receives a very limited information about the context in which it was initialized (only the JobID). 

For more elaborate state backends (for instance db backends) it becomes necessary to have more information about the running job to allow more fine-grained handling or sharding of the stored state. 

Changes: 
 - The initalizeForJob method should receive the Environment of the task from which it was called instead of just the jobId. This allows getting information about parallelism and subtask index for instance.
- The createKvState method should additionally receive the unique id of the operator in the jobgraph along with the name of the created state name (jobid + operatorId + statename creates a unique identifier for the kvState)





--
This message was sent by Atlassian JIRA
(v6.3.4#6332)