Answer-first summary for fast verification
Answer: Network I/O never spikes, Overall cluster CPU utilization is around 25%
The question asks which Ganglia Metrics indicate a bottleneck caused by code executing on the driver. When the driver is overloaded (e.g., due to serial processing or task scheduling delays), the executors remain underutilized. - **Option E**: Overall cluster CPU utilization around 25% suggests that executors are idle (since the driver's high CPU usage, when averaged across all nodes, results in low overall utilization). - **Option D**: Network I/O never spikes because the driver is not efficiently distributing tasks to executors, leading to minimal data exchange. Other options like A, B, and C relate to load averages, disk space, or network bytes, which are less directly tied to driver-specific bottlenecks in this scenario.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
In a production cluster with 3 executor nodes where the driver and executors use the same VM type, which Ganglia Metrics indicator would identify a performance bottleneck due to driver-side code execution?
A
The five Minute Load Average remains consistent/flat
B
Bytes Received never exceeds 80 million bytes per second
C
Total Disk Space remains constant
D
Network I/O never spikes
E
Overall cluster CPU utilization is around 25%
No comments yet.