[STORM-2298] Don't kill Nimbus when ClusterMetricsConsumer is failed to initialize

• Type: Improvement
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.0.0, 1.1.0
• Fix Version/s: 2.0.0, 1.1.0
• Component/s: storm-core
• Epic Link: Release Apache Storm 1.1.0
• Assignee: Jungtaek Lim
• Reporter: Jungtaek Lim
• Created: 17/Jan/17 14:04
• Updated: 24/Jan/17 03:44
• Resolved: 24/Jan/17 03:44

Description

ClusterMetricsConsumerExecutor doesn't pass errors to Nimbus when sending metrics to ClusterMetricsConsumer, but passes errors to Nimbus when initializing ClusterMetricsConsumer and kills Nimbus.

Nimbus should have fault tolerance on ClusterMetricsConsumer so that it could drop cluster metrics but still be alive.

It might be also ideal to retry initialization if ClusterMetricsConsumerExecutor is about to send metrics but initialization of ClusterMetricsConsumer is failed.

Issue Links

• links to: GitHub Pull Request #1881