[ 
https://issues.apache.org/jira/browse/OAK-3305?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14716483#comment-14716483
 ] 

Marcel Reutegger commented on OAK-3305:
---------------------------------------

The issue seems to be caused by the sequence of operations performed in the 
DocumentNodeStore constructor.

# head revision is set
# root state is obtained
# _lastRev recovery is performed

The last step changes the state of the root document while the head revision is 
still the same. The last two steps should probably be switched.

> Self recovering instance may not see all changes
> ------------------------------------------------
>
>                 Key: OAK-3305
>                 URL: https://issues.apache.org/jira/browse/OAK-3305
>             Project: Jackrabbit Oak
>          Issue Type: Bug
>          Components: core, mongomk
>    Affects Versions: 1.0, 1.2, 1.3.4
>            Reporter: Marcel Reutegger
>            Assignee: Marcel Reutegger
>              Labels: resilience
>             Fix For: 1.3.6
>
>
> When a DocumentNodeStore instance is killed and restarted, the _lastRev 
> recovery mechanism is triggered on startup. It may happen that the restarted 
> instance does not see all changes that were recovered.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to