[SAMZA-1150] Handling Error propagation between ZkJobCoordinator & DebounceTimer

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.13.0
• Assignee: Navina Ramesh
• Reporter: Navina Ramesh
• Created: 17/Mar/17 01:18
• Updated: 09/May/17 00:59
• Resolved: 09/May/17 00:59

Description

Right now, I don't think we handle errors from any of the callback properly. As far as I can tell, most of the errors should cause the processor to stop and propagate the error to the user. But we need to carefully evaluate the various error scenarios.
Through this JIRA, we need to answer the following:
1. How to handle errors/exceptions thrown in the callbacks that are executed with the debounce timer?
2. Debounce timer has a category of tasks to run. What happens when one task fails? Do we cancel the other pending tasks, if any ? Does the ordering of the tasks matter at all?
3. What kind of failures should be considered FATAL?

Issue Links

• is depended upon by: SAMZA-1064 Standalone Samza with Zookeeper for Coordination
• links to: GitHub Pull Request #166

Sub-Tasks

• Leader should hold a lock to avoid failures on long GC pause/timeouts: Open
• Remove DebounceTimer dependency from ZkLeaderElector: Resolved
• Modify schedule api in ScheduleAfterDebounceTime to take an error handling callback: Open