[FLINK-4156] Job with -m yarn-cluster registers TaskManagers to another running Yarn session

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.1.0
• Component/s: Deployment / YARN, Runtime / Coordination
• Assignee: Unassigned
• Reporter: Stefan Richter
• Created: 06/Jul/16 08:37
• Updated: 19/Dec/16 12:46
• Resolved: 26/Jul/16 16:18

Description

When a job is started using cluster mode (-m yarn-cluster) and a Yarn session is running on the same cluster, the job accidentally registers it's worker tasks with the ongoing Yarn session. This happens because the same Zookeeper namespace is used.

We should consider isolating Flink applications from another by using UUIDS, e.g. based on their application ids, in their Zookeeper paths.

Issue Links

• relates to: FLINK-4166 Generate automatic different namespaces in Zookeeper for Flink applications