Xuze Yang created HDFS-17077:
--------------------------------
Summary: slow disks should be used when no more normal disks are
available
Key: HDFS-17077
URL: https://issues.apache.org/jira/browse/HDFS-17077
Project: Hadoop HDFS
Issue Type: Improvement
Affects Versions: 3.3.5
Reporter: Xuze Yang
Attachments: image-2023-07-10-15-12-21-760.png
When slow disks detection is enabled, hdfs can distinguish __ slow disks among
all disks in a datanode. When a disk is regard as a slow disk, it will be
excluded when choosing volume. The code is shown as below:
!image-2023-07-10-15-12-21-760.png!
Problem appears if all normal disks is fulled or being fatal state while slow
disks is available, then no disks will be chosen and no data can be
successfully written. In this situation, I think we should let slow disks be
chosen, it's more reasonable.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org
