[FLINK-9949] Jepsen: Kill Flink processes when tearing down cluster

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 1.6.0
• Fix Version/s: 1.6.0
• Component/s: Tests
• Labels: pull-request-available
• Assignee: Gary Yao
• Reporter: Gary Yao
• Created: 25/Jul/18 13:21
• Updated: 25/Jul/18 16:44
• Resolved: 25/Jul/18 15:40

Description

Because Flink processes are not killed at the end of the tests, it can happen that surviving instances create znodes in the ZooKeeper ensemble of the next test run. This creates ambiguity when we retrieve the address of the leading REST server from ZooKeeper.

Issue Links

• links to: GitHub Pull Request #6419
• links to: GitHub Pull Request #6419