[
https://issues.apache.org/jira/browse/HDDS-4738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17335448#comment-17335448
]
Mark Gui commented on HDDS-4738:
--------------------------------
IMHO, 'no healthy' should be classified as at least 2 kinds:
* bad disks.
* good disks with no space left(but still readable).
Only a datanode with all its disks detected as bad disks, should be DEAD(or
DECOMMISSIONED ?), or even if it only has 1 disk which is good(maybe no space
left) should be in a HEALTHY_READONLY state, then this datanode could still
serve read requests but do not join any pipeline.
(There is a similar NodeState definition in an issue, not directly related to
disk issues, but defines that state:
https://issues.apache.org/jira/browse/HDDS-4992)
We shall handle the following conditions:
* all bad disks -> DEAD / DECOMMISSIONED
* all nospace disks -> HEALTHY_READONLY
* some bad + some nospace. -> HEALTHY_READONLY
* at least one good -> normal case
And there are 2 kinds of disks:
* data disks
* meta disks
Data disks should have the conditions above, and for meta disks which are used
by ratis, we could only be notified if there are bad/nospace disks. (Should
rocksdb be on a separate disk?)
[~shashikant] What do you think?
> Mark Datanode with no heathy data or metadata disks as dead in SCM
> ------------------------------------------------------------------
>
> Key: HDDS-4738
> URL: https://issues.apache.org/jira/browse/HDDS-4738
> Project: Apache Ozone
> Issue Type: Sub-task
> Components: SCM
> Reporter: Shashikant Banerjee
> Assignee: Shashikant Banerjee
> Priority: Major
>
> If a datanode reports no healthy data disks or metadata disks, it should not
> participate in any new pipeline allocation strategy. The idea here is to mark
> such nodes as DEAD.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]