[
https://issues.apache.org/jira/browse/HDFS-9422?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15022007#comment-15022007
]
He Liangliang commented on HDFS-9422:
-
Just record the time consumed, jstack not captured. ~5s(mean
[
https://issues.apache.org/jira/browse/HDFS-9422?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
He Liangliang updated HDFS-9422:
Description: In DirectoryScanner, the scan call hold the global dataset
lock for quite long time
[
https://issues.apache.org/jira/browse/HDFS-9422?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15007908#comment-15007908
]
He Liangliang commented on HDFS-9422:
-
No, the lock can make NO difference, EXCEPT we move the
[
https://issues.apache.org/jira/browse/HDFS-9422?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
He Liangliang updated HDFS-9422:
Attachment: HDFS-9422.patch
The lock is unnecessary:
1. getFinalizedBlocks is thread safe and the
He Liangliang created HDFS-9422:
---
Summary: Unnecessary FsDatasetImpl locking in DirectoryScanner
cause periodic datanode pauses
Key: HDFS-9422
URL: https://issues.apache.org/jira/browse/HDFS-9422
