[
https://issues.apache.org/jira/browse/HDDS-12169?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ryan Blough closed HDDS-12169.
------------------------------
Assignee: Ryan Blough
Closed as duplicate.
> Add a RackScatter container placement policy for RATIS
> ------------------------------------------------------
>
> Key: HDDS-12169
> URL: https://issues.apache.org/jira/browse/HDDS-12169
> Project: Apache Ozone
> Issue Type: New Feature
> Affects Versions: 1.4.0
> Reporter: Ryan Blough
> Assignee: Ryan Blough
> Priority: Minor
>
> I observed a scenario where a commercial Ozone user had rack awareness
> configured, using standard RATIS 3 containers, and they experienced data loss
> due to a series of unfortunate events.
> The key bad event for them is that their hardware was managed by a vendor,
> who shut down one of their racks for maintenance by surprise. This took 2 out
> of 3 replicas down for long enough to trigger replication, and in multiple
> cases the remaining containers had been impacted by the opened-twice because
> of volume failure problem, resulting ultimately in data loss.
> It seems like this problem would have been averted if the loss of a single
> rack didn't make a majority of replicas unavailable.
> I believe the solution is to provide a container placement policy for RATIS
> like the RackScatter policy for EC.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
