[ 
https://issues.apache.org/jira/browse/OAK-5557?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Stefan Egli updated OAK-5557:
-----------------------------
    Attachment: DocumentNodeStoreBranchesTest.java

Second, slightly simpler variant of [^DocumentNodeStoreBranchesTest.java] 
attached which consistently reproduces the problem (tested on 1.2.14 only for 
now). The sequence of critical events seem to be:
* (A) a number of purges in 5-sec window A
* (B) a merge from an unrelated session in 5-sec window A+1
* the merge of the (A) changes then result in diffManyChildren to use as 
{{minValue}} window A+1 instead of A. This is likely due to the rebase

> incomplete diffManyChildren during commitHook evaluation in a persisted branch
> ------------------------------------------------------------------------------
>
>                 Key: OAK-5557
>                 URL: https://issues.apache.org/jira/browse/OAK-5557
>             Project: Jackrabbit Oak
>          Issue Type: Bug
>    Affects Versions: 1.2.14, 1.2.23
>         Environment: 'easily' reproducible on 1.2.14
> 'takes longer to' reproduce on 1.2.23
>            Reporter: Stefan Egli
>            Assignee: Stefan Egli
>         Attachments: DocumentNodeStoreBranchesTest.java, 
> DocumentNodeStoreBranchesTest.java
>
>
> In oak 1.2 when doing a diffManyChildren as part of running the CommitHooks 
> as part of a branch commit, not all child paths are taken into account in 
> some cases.
> The problem seems to be related to the [{{_modified}} 
> calculation|https://github.com/apache/jackrabbit-oak/blob/jackrabbit-oak-1.2.14/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/document/DocumentNodeStore.java#L2242]
>  when doing the query for affected child paths. It seems [that 
> {{minValue}}|https://github.com/apache/jackrabbit-oak/blob/jackrabbit-oak-1.2.14/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/document/DocumentNodeStore.java#L2242]
>  is not correctly calculated when doing {{compareAgainstBaseState}} as part 
> of a persisted branch.
> Note that the problem can be reproduced very good on 1.2.14, less good but 
> still on the latest 1.2 and not in trunk (ie almost 1.6.0).
> Attaching a test case to reproduce



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to