[HDFS-10475] Adding metrics for long FSNamesystem read and write locks

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: namenode
• Labels: metrics
• Target Version/s: 2.7.8
• Assignee: Erik Krogen
• Reporter: Xiaoyu Yao
• Created: 01/Jun/16 18:38
• Updated: 04/Jun/18 17:37

Description

This is a follow up of the comment on HADOOP-12916 and here add more metrics and WARN/DEBUG logs for long FSD/FSN locking operations on namenode similar to what we have for slow write/network WARN/metrics on datanode.

Issue Links

• is related to: HDFS-10713 Throttle FsNameSystem lock warnings
• is related to: HDFS-9145 Tracking methods that hold FSNamesystemLock for too long
• is related to: HDFS-10742 Measure lock time in FsDatasetImpl
• is related to: HDFS-10817 Add Logging for Long-held NN Read Locks
• relates to: HADOOP-13521 SampleQuantile does not perform well under load

Sub-Tasks

• Add histograms for FSNamesystemLock Metrics: Open
• Add FSNamesystemLock metrics for BlockManager operations: Open