[
https://issues.apache.org/jira/browse/HDDS-3952?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17159636#comment-17159636
]
runzhiwang edited comment on HDDS-3952 at 7/17/20, 3:01 AM:
------------------------------------------------------------
[~msingh] Hi, Thanks for review. The backgroud of this jira is our production
cluster has about 7000 small containers which are not full but closed, as the
image shows, because ratis pipeline is not stable. So we want to reopen and
write to the closed but not full container.
The basic idea:
1. SCM build a map with entry <3 datanodes, set of closed but not full
containers on the 3 datanodes>
2. SCM check whether any open pipeline locate on the 3 datanodes from the
map.entrySet, if exists such open pipeline, we get the closed but not full
containers by map.get(pipeline.datanodes()), put them on the pipeline and
reopen them.
3. When SCM create new pipeline, we first select from the map which 3 datanodes
has the most closed but not full containers, and create pipeline on this 3
datanodes.
[~msingh] What do you think ?
!screenshot-1.png!
was (Author: yjxxtd):
[~msingh] Hi, Thanks for review. The backgroud of this jira is our production
cluster has about 7000 small containers which are not full but closed, as the
image shows, because ratis pipeline is not stable. So We want to reopen and
write to the closed but not full container.
The basic idea:
1. SCM build a map with entry <3 datanodes, set of closed but not full
containers on the 3 datanodes>
2. SCM check whether any open pipeline locate on the 3 datanodes from the
map.entrySet, if exists such open pipeline, we get the closed but not full
containers by map.get(pipeline.datanodes()), put them on the pipeline and
reopen them.
3. When SCM create new pipeline, we first select from the map which 3 datanodes
has the most closed but not full containers, and create pipeline on this 3
datanodes.
[~msingh] What do you think ?
!screenshot-1.png!
> Make container could be reopened
> --------------------------------
>
> Key: HDDS-3952
> URL: https://issues.apache.org/jira/browse/HDDS-3952
> Project: Hadoop Distributed Data Store
> Issue Type: Improvement
> Components: Ozone Datanode, SCM
> Affects Versions: 0.7.0
> Reporter: maobaolong
> Assignee: runzhiwang
> Priority: Major
> Attachments: screenshot-1.png
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: ozone-issues-h...@hadoop.apache.org
