Search before asking
Description
Now when Master run a workflow, it will do many db operation, which will make Master handle slowly.
Such as a easy dag, contains a shell task, run once and will occur more than 30 db operation: 1 delete, 2 insert, 5 update and 26 select.
I think if Master use memory cache of processInstance and taskInstance, instead of selecting from db every time, it will reduce the db stress and improve the Master handling workflow.
- when api update process instance or task instance, notify master to refresh and change state.
WorkflowExecuteThread as the cache manager for each processInstance, which contains the taskInstances of this workflow.
BTW, I found that the old design TaskInstanceCacheManagerImpl, but it seems never used now.
Use case
No response
Related issues
No response
Are you willing to submit a PR?
Code of Conduct
Search before asking
Description
Now when Master run a workflow, it will do many db operation, which will make Master handle slowly.
Such as a easy dag, contains a shell task, run once and will occur more than 30 db operation: 1 delete, 2 insert, 5 update and 26 select.
I think if Master use memory cache of processInstance and taskInstance, instead of selecting from db every time, it will reduce the db stress and improve the Master handling workflow.
WorkflowExecuteThreadas the cache manager for each processInstance, which contains the taskInstances of this workflow.BTW, I found that the old design
TaskInstanceCacheManagerImpl, but it seems never used now.Use case
No response
Related issues
No response
Are you willing to submit a PR?
Code of Conduct