[
https://issues.apache.org/jira/browse/RATIS-644?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16899082#comment-16899082
]
Bharat Viswanadham edited comment on RATIS-644 at 8/2/19 5:52 PM:
------------------------------------------------------------------
Hi [~sdeka]
{quote}Instead, the desired behaviour is to stop accepting new append requests
on the group
{quote}
For OM case, when any transaction is failed, we should fail and stop that OM.
For OM case stopping ratis server might be the correct way to do.
If we continue, OM DB's might get out of sync.
was (Author: bharatviswa):
Hi [~sdeka]
{quote}Instead, the desired behaviour is to stop accepting new append requests
on the group
{quote}
For OM case, when any transaction is failed, we should fail and stop that OM.
For OM case stopping ratis server might be the correct way to do.
> Handle failures encountered in applyLog
> ---------------------------------------
>
> Key: RATIS-644
> URL: https://issues.apache.org/jira/browse/RATIS-644
> Project: Ratis
> Issue Type: Bug
> Components: server
> Reporter: Supratim Deka
> Assignee: Supratim Deka
> Priority: Major
>
> This is related to HDDS-1739 which detects and reports IO failures during
> applyTransaction.
> Currently, the StateMachineUpdater in response to a failed
> applyLog/applyTransaction operation triggers a shutdown of the Ratis server.
> Instead, the desired behaviour is to stop accepting new append requests on
> the group. For the case of Ozone, the state machine will initiate
> destruction/closing of the Ratis group - this is implemented as a part of
> HDDS-1739.
>
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
