[
https://issues.apache.org/jira/browse/OAK-3455?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14943071#comment-14943071
]
Marcel Reutegger commented on OAK-3455:
---------------------------------------
I didn't intend to change anything else than the log message, which should not
be interpreted by code to make a decision whether a retry should be done.
OAK-3042 already improved conflict handling across cluster nodes and suspends
the commit until the conflicting change from another cluster node is visible.
This at least ensures the conflict is visible after the
InvalidItemStateException, which was not the case before. But even with
OAK-3042 there is no guarantee a retry on the application level will succeed.
With every retry there is a chance of a conflict change done on another cluster
node. Conflicts on a local cluster node can be handled differently, because the
DocumentNodeStore will internally retry the commit with an exclusive lock and
effectively serialize commits (on the local cluster node). There is no such
mechanism across cluster nodes.
> Improve conflict exception message
> ----------------------------------
>
> Key: OAK-3455
> URL: https://issues.apache.org/jira/browse/OAK-3455
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: core, mongomk
> Reporter: Marcel Reutegger
> Assignee: Marcel Reutegger
> Priority: Minor
>
> The conflict exception message for the DocumentNodeStore could be improved by
> adding information about the visibility of the conflict revision.
> A conflict across cluster nodes is the most likely case and it would be good
> to know if the other revision is already visible to the current cluster node.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
