[
https://jira.nuxeo.com/browse/NXP-7127?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Florent Guillaume resolved NXP-7127.
------------------------------------
Resolution: Won't Fix
Not possible to fix, concurrent access to a document's complex properties
should be avoided.
> Wrong storage of complex lists in concurrent environment
> --------------------------------------------------------
>
> Key: NXP-7127
> URL: https://jira.nuxeo.com/browse/NXP-7127
> Project: Nuxeo Enterprise Platform
> Issue Type: Bug
> Components: Core SQL Storage
> Affects Versions: 5.3 GA
> Environment: Ubuntu 10.04
> Nuxeo-5.4.2
> mvn -Dtest=TestConcurrentRepositoryAPI test
> Reporter: Alexander Litvin
> Assignee: Florent Guillaume
> Priority: Critical
> Fix For: 5.5
>
> Attachments: select-for-update.diff,
> TestConcurrentRepositoryAPI.java, TestConcurrentRepositoryAPI.java
>
>
> For lists of complex properties (vignettes is the list of vignette elements
> in the provided test) we can see that in a concurrent environment (several
> users updating the same document with the same data (a list of two elements
> in our case)) Nuxeo stores an incorrect collection (the amount of vignettes
> become more than user actually stored). It is reproducible for properties
> stored as separate records in hierarchy (not for simple collections like the
> list of strings). The amount of stored vignettes is not predictable,
> sometimes it become 4 instead of expected 2, sometimes it is 8 or any other
> number which is bigger than 2.
> Background: I wrote this test after we started experiencing the following
> problem in our development environment: sometimes when user stored some
> document changing just the title, after saving the collections of complex
> properties in this document would double. The Issue becomes reproducible more
> often if the user performs a save of the same document often one by one even
> manually.
> Steps to reproduce: In order to reproduce the issue you can run the attached
> test (put it into nuxeo-core-storage-sql-test module). The issue could not be
> reproducible in some circumstances, it depends on machine configuration, in
> this case you should change THREAD_TIMEOUT in testParallelUpdateComplexList
> method until the tests stop passing.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
_______________________________________________
ECM-tickets mailing list
[email protected]
http://lists.nuxeo.com/mailman/listinfo/ecm-tickets
