[jira] [Updated] (OAK-1420) ConcurrentAddIT fails on buildbot

["Michael Dürig (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/OAK-1420?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Michael Dürig updated OAK-1420:
-------------------------------

    Attachment: OAK-1420.log
                OAK-1420.patch

With this patch I was able to reproduce the problem locally. Increase {{org.apache.jackrabbit.oak.jcr.ConcurrentAddIT#NODES_PER_WORKER}} if needed. 

The patch also dumps the name of the missing nodes as well as all merges from {{org.apache.jackrabbit.oak.plugins.commit.ChildOrderConflictHandler#merge}}. 

In one case I observed the test to fail with {{worker-20-node-0}} and {{worker-20-node-1}} missing. Looking this up in the merges I found:

{code}
ours: worker-20-node-0, worker-20-node-1, 
theirs: worker-17-node-0, 
merged: worker-17-node-0, worker-20-node-0, worker-20-node-1, 
{code}

So there was a conflict, which was successfully and correctly merged. However it seems that the result of this merge gets somehow overwritten later on as neither {{worker-20-node-0}} nor {{worker-20-node-1}} appear in any of the many later merges. 

See OAK-1420.log for the full console dump of this run.

> ConcurrentAddIT fails on buildbot
> ---------------------------------
>
>                 Key: OAK-1420
>                 URL: https://issues.apache.org/jira/browse/OAK-1420
>             Project: Jackrabbit Oak
>          Issue Type: Bug
>          Components: mongomk
>            Reporter: Michael Dürig
>         Attachments: OAK-1420.log, OAK-1420.patch
>
>
> {{org.apache.jackrabbit.oak.jcr.ConcurrentAddIT#addNodesSameParent}} currently fails consistently on buildbot for the document NS fixture: https://ci.apache.org/builders/oak-trunk/builds/4375. It does not fail locally for me though. 



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)