[
https://issues.apache.org/jira/browse/HDDS-11328?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17875675#comment-17875675
]
Duong commented on HDDS-11328:
------------------------------
This is implemented together with HDDS-11239.
> ContainerStateMachine should not crash because of CHUNK_FILE_INCONSISTENCY
> --------------------------------------------------------------------------
>
> Key: HDDS-11328
> URL: https://issues.apache.org/jira/browse/HDDS-11328
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: Duong
> Assignee: Duong
> Priority: Major
> Fix For: 1.5.0
>
> Attachments: Screenshot 2024-08-16 at 10.46.06 AM.png
>
>
> Since HDDS-9844, we allow clients to send writeChunks without waiting for the
> last writeChunk.
> This leads to the possibility that a datanode can receive an off-order chunk
> due to a temporary network error.
> !Screenshot 2024-08-16 at 10.46.06 AM.png|width=718,height=185!
> Today, when that happens, the pipeline crashes and keeps crashing after
> restarts and leader changes.
> Off-order writeChunks should be rejected logically as an expected error code,
> just like CONTAINER_NOT_OPEN and CLOSED_CONTAINER_IO so that clients can
> retry on another block/container/pipeline.
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
