tomscut created HDFS-16505:
------------------------------
Summary: Setting safemode should not be interrupted by abnormal
nodes
Key: HDFS-16505
URL: https://issues.apache.org/jira/browse/HDFS-16505
Project: Hadoop HDFS
Issue Type: Bug
Reporter: tomscut
Assignee: tomscut
Attachments: image-2022-03-15-09-29-36-538.png,
image-2022-03-15-09-29-44-430.png
Setting safemode should not be interrupted by abnormal nodes.
For example, we have four namenodes configured in the following order:
NS1 -> active
NS2 -> standby
NS3 -> observer
NS4 -> observer.
When the {color:#FF0000}NS1 {color}process exits, setting the states of
safemode, {color:#FF0000}NS2{color}, {color:#FF0000}NS3{color}, and
{color:#FF0000}NS4 {color}fails. Similarly, when the {color:#FF0000}NS2{color}
process exits, only the safemode state of {color:#FF0000}NS1{color} can be set
successfully.
When the {color:#FF0000}NS1{color} process exits:
Before the change:
!image-2022-03-15-09-29-36-538.png|width=1145,height=97!
After the change:
!image-2022-03-15-09-29-44-430.png|width=1104,height=119!
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org
