[
https://issues.apache.org/jira/browse/HDDS-14083?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Siddhant Sangwan updated HDDS-14083:
------------------------------------
Description:
In a Ratis pipeline of three Datanodes, suppose the leader's data disk has
reached full logical capacity (ozone capacity, ie, raw capacity - du.reserved -
min.free.space). If chunk write is made to this pipeline, the container will
not be created on the leader DN. However it will still be created on the
follower DNs because state machine data write and log write is done in parallel
for all DNs.
This will cause empty QUASI_CLOSED containers to be created in this cluster.
These containers will stay stuck in the system. This was discovered while
testing, no action is planned for these containers as of now.
was:
In a Ratis pipeline of three Datanodes, suppose the leader's data disk has
reached full logical capacity (ozone capacity, ie, raw capacity - du.reserved -
min.free.space). If chunk write is made to this pipeline, the container will
not be created on the leader DN. However it will still be created on the
follower DNs because state machine data write and log write is done in parallel
for all DNs.
This will cause empty QUASI_CLOSED containers to be created in this cluster.
> Empty QUASI_CLOSED containers created when leader DN's disk is full
> -------------------------------------------------------------------
>
> Key: HDDS-14083
> URL: https://issues.apache.org/jira/browse/HDDS-14083
> Project: Apache Ozone
> Issue Type: Sub-task
> Components: Ozone Datanode
> Affects Versions: 2.1.0
> Reporter: Siddhant Sangwan
> Assignee: Siddhant Sangwan
> Priority: Major
>
> In a Ratis pipeline of three Datanodes, suppose the leader's data disk has
> reached full logical capacity (ozone capacity, ie, raw capacity - du.reserved
> - min.free.space). If chunk write is made to this pipeline, the container
> will not be created on the leader DN. However it will still be created on the
> follower DNs because state machine data write and log write is done in
> parallel for all DNs.
> This will cause empty QUASI_CLOSED containers to be created in this cluster.
> These containers will stay stuck in the system. This was discovered while
> testing, no action is planned for these containers as of now.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
