YongGang commented on PR #15976: URL: https://github.com/apache/druid/pull/15976#issuecomment-1970321082
> Instead of the new getLabel method in Task, we should have something more like getTaskLaneInfo that returns a TaskLaneInfo object which may be augmented later if needed. It should also be possible to manipulate the details of the lane info (i.e. the label in this case) using the task context. I guess the current problem in scheduling arises from the fact that we are limited to using immutable fields such as type and datasource (and maybe context parameter taskPriority) to determine how a task should be scheduled. Powering this through the task context will allow users much more flexibility. It's a good idea. However, I'd like to emphasize that task labeling serves multiple purposes beyond scheduling. For instance, identifying the costs associated with querying from deep storage tasks can be hard now, as they fall under the same MSQ task type as others. With task labeling, we can include this information in task reports, facilitating detailed cost analysis. Therefore, I propose keeping the `getLabel` method separate, enabling its use in various contexts beyond scheduling alone. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
