[HDFS-14628] Improve information on scanAndCompactStorages in BlockManager and lower log level

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 3.1.0
• Assignee: Shen Yinjie
• Reporter: Shen Yinjie
• Created: 03/Jul/19 04:28
• Updated: 03/Jul/19 05:49
• Resolved: 03/Jul/19 05:47

Description

BlockManager#scanAndCompactStorages is called every 600sec by default.
In big cluster with thousands of datanodes, it will print out 10 thousands of informations every sec when scanAndCompactStorages() is running,which may make much noise in namenode logs. And, currently these INFO logs at namenode side and does not provide much information.

Issue Links

• is duplicated by: HDFS-13692 StorageInfoDefragmenter floods log when compacting StorageInfo TreeSet