nishita-09 commented on PR #1000:
URL: 
https://github.com/apache/flink-kubernetes-operator/pull/1000#issuecomment-4212987894

   > https://github.com/apache/flink-kubernetes-operator/pull/1080/changes
   
   
   
   > I agree that this manual mapping of configs is very fragile , and it 
duplicates a lot of logic from core Flink. A different approach was taken in 
https://github.com/apache/flink-kubernetes-operator/pull/1080/changes it uses a 
different endpoint to get the JobManagerJobConfiguration.
   > 
   > Can we simply use this instead of all runtimeConfig/checkpointConfig ? I 
think when the user programmatically sets a config in the StreamExecEnv those 
actually end up in the JobManagerJobConfiguration. cc @Dennis-Mircea
   
   @gyfora For checkpoints configurations , the programmatically set config 
does not end up in JobManagerJobConfiguration. This still shows the default 
configs set in the deployment, while the checkpoint/config endpoint is the 
place where the configurations correctly reflect.
   
   When user code calls `env.enableCheckpointing(30000)`, it writes 
`CheckpointingOptions.CHECKPOINTING_INTERVAL = 30s` into the Configuration 
object inside StreamExecutionEnvironment. This is the same Configuration object 
that gets serialized into the JobGraph and eventually becomes the 
CheckpointCoordinatorConfiguration used by the CheckpointCoordinator at runtime.
   
   CheckpointConfigHandler - This handler serves 
`/jobs/:jobid/checkpoints/config`. It reads from the execution graph, which 
gets its values from the CheckpointCoordinatorConfiguration, which was built 
from the JobGraph (which includes the programmatic overrides)
   
   JobManagerJobConfigurationHander - This handler serves` 
/jobs/:jobid/jobmanager/config`. It reads from a pre-baked Configuration object 
passed at construction time i.e the JM's initial configuration.
   The Configuration object passed to this handler is the JM's original startup 
Configuration. The user's main() method hasn't run yet when this handler is 
constructed, and even after it runs, it only modifies the 
StreamExecutionEnvironment's internal Configuration and not the JM-level 
Configuration. 
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to