[Feature][all] We are ready to contribute multiple features
liukunyuan opened this issue · comments
Search before asking
- I had searched in the issues and found no similar feature requirement.
Description
We have these feature within our company and hope to contribute them. What feature communities require?
- Worker and master grayscale function based on traffic
- Optimization of worker and master load protection strategies (optimizing remaining physical memory and CPU algorithms on the machine) to address the issue of worker downtime
- Force skip dependency detection during complement
- Heartbeat reporting from master and worker to detect disconnection issues
- Write the Yarn appid into the task_instance table during task execution to facilitate worker fault tolerance (solving the problem of worker machine downtime and inability to tolerate faults)
- Workflow migration project interface
- Recommend users to set the cron for the workflow (query the completion time of all upstream workflows in the last 7 days and then recommend the start time of this workflow)
- After the task timeout fails, it is necessary to kill the Yarn task
- Change workflow instances and task instances to single table queries to solve the problem of slow queries
- Provide a batch recovery failure interface to solve the problem of slow recovery of workflow instances caused by cluster or MySQL issues
- Add feature control tables to MySQL. All important functions require parameter control to turn on and off
Are you willing to submit a PR?
- Yes I am willing to submit a PR!
Code of Conduct
- I agree to follow this project's Code of Conduct
What causes the fourth problem?
And I have some suggest:
1.The fifth problem may cause task loss. A data processing task may contain more then one yarn job. If it fails between yarn tasks, this may cause the fault-tolerant task to succeed but miss some yarn task submissions.
2.Can the eighth one be configured? The easy failure of some long tasks may lead to a waste of complement resources.
Please Create DSIP to describe why you want to do this change, it's hard to know what you want to do by a simple description, and some of these should already been supported.
What causes the fourth problem? And I have some suggest: 1.The fifth problem may cause task loss. A data processing task may contain more then one yarn job. If it fails between yarn tasks, this may cause the fault-tolerant task to succeed but miss some yarn task submissions. 2.Can the eighth one be configured? The easy failure of some long tasks may lead to a waste of complement resources.
- What causes the fourth problem?
Mysql network exception for a few minutes - A data processing task may contain more then one yarn job。
It will write multiple appids to the task_instance table