[HADOOP-5031] metrics aggregation is incorrect in database

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.21.0

Environment

Redhat 5.1, Java 6

• Hadoop Flags: Reviewed
• Assignee: Eric Yang
• Reporter: Eric Yang
• Created: 14/Jan/09 19:01
• Updated: 24/Aug/10 20:34
• Resolved: 06/Mar/09 18:13

Description

A few problem with the aggregation SQL statements:

hdfs throughput should be calculated by doing two level aggregation:

First, calculate the rate for hadoop datanode metrics with accumulated vales.
Second, sum up all datanode rate to provide a single number to represent the current cluster performance.

Disable hod jobs utilization measurement - The data provide a rough view of the cluster performance but mostly inaccurate.
Disable user utilization measurement generated from hod job - The data is generated from hod job metrics, and it's mostly inaccurate.