Details
-
Feature Request
-
Resolution: Unresolved
-
L3 - Default
-
None
-
None
-
None
Description
User Story (Required on creation):
When downscaling Camunda instances in a cluster, the job executor may shut down while it has already acquired jobs for execution. These jobs should be unlocked so that they become immediately available for execution again, not only after the job lock expires.
Functional Requirements (Required before implementation):
- After a Camunda instance is shut down normally, there are no more locked jobs by the corresponding job executor in the database
Technical Requirements (Required before implementation):
- Ideally this unlocking happens after the thread pool has shut down to reduce the risk of multiple executions of the same job to a minimum. However, this may be hard to implement given that we use different thread pools in the different application server setups and there may be no concepts to hook into thread pool shutdown
Limitations of Scope (Optional):
Hints (Optional):
Option 1: Before the acquisition thread terminates, unlock all jobs of its lock owner (if this is not synchronized with the shutdown of the threadpool, it may result in jobs being executed multiple times)
Option 2: When the thread pool is shut down, drain the current job executor queue and unlock all elements (this may be hard to achieve, because we probably do not have access to the queue in all job executor setups)
mgm-controller-panel
This is the controller panel for Smart Panels app
Attachments
Issue Links
- mentioned in
-
Page Loading...