[
https://issues.apache.org/jira/browse/HDDS-494?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16619128#comment-16619128
]
Shashikant Banerjee commented on HDDS-494:
------------------------------------------
Through SCMCLI, we call closeContainer on the Datanodes first and then move the
container state to CLOSING state though ideally we the container should be
moved to CLOSING state in SCM first followed by execution of CLOSE container on
Datanode. Until the closeContainer command completes on the Datanode, new
blocks can still get allocated from the containers which should have been
removed from the allocation path by this time. CloseContainer command also
should only be executed via ratis not via standalone servers which may lead to
inconsistency in the closeContainer replica.
> client fails to write when all the containers are closed through scmcli
> ------------------------------------------------------------------------
>
> Key: HDDS-494
> URL: https://issues.apache.org/jira/browse/HDDS-494
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Affects Versions: 0.2.1
> Reporter: Nilotpal Nandi
> Assignee: Shashikant Banerjee
> Priority: Major
> Attachments: all-node-ozone-logs-1537191049.tar.gz
>
>
> steps taken :
> ------------------
> # running putKey from cli.
> # closed all containers.
> # Client fails to write and throws error on cli (related to bug HDDS-482 )
>
> Expectation :
> -------------------
> there should be retry and allocate new containers .
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
