[
https://issues.apache.org/jira/browse/HDDS-9197?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Attila Doroszlai updated HDDS-9197:
-----------------------------------
Fix Version/s: 1.4.0
> EC: Mark EC containers unhealthy when not missing but unrecoverable
> -------------------------------------------------------------------
>
> Key: HDDS-9197
> URL: https://issues.apache.org/jira/browse/HDDS-9197
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: Stephen O'Donnell
> Assignee: Stephen O'Donnell
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.4.0
>
>
> An EC container should only be marked as unhealthy if:
> 1) Is not missing (has at least dataNum replica indexs online)
> 2) Has at least dataNum Healthy indexes online.
> For a 3-2 container, this means that:
> Healthy, Healthy -> This is missing
> Unhealthy, Unhealthy -> This is missing as it doesn't matter that the
> replicas are unhealthy, as we could not read it anyway as we need at least 3
> replicas.
> Healthy, Healthy, Unhealthy (indexes 1, 2, 3) -> This is unhealthy as
> technically much of the data can still be read, but it is the unhealthy
> container which is causing the problem and making it unrecoverable.
> Healthy, Healthy, Healthy, Unhealthy -> This is under replicated, as we can
> recover using the 3 healthy.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
