Catalog corruption when multiple admins hack the configuration
--------------------------------------------------------------
Key: GEOS-2408
URL: http://jira.codehaus.org/browse/GEOS-2408
Project: GeoServer
Issue Type: New Feature
Components: Configuration
Affects Versions: 1.7.0
Reporter: Andrea Aime
Assignee: Andrea Aime
Priority: Critical
Fix For: 1.7.1
A user is reporting a situation in which the catalog gets currupted quite often, as a result of multiple admins playing with GeoServer at the same time.
In particular with have a few human users and the GeoSolution ingestion engine, but we could have the RESTConfig module in the play as well (I mean, in theory).
I looked around, whilst most of the Catalog information access is synchonized, the apply and save actions are globally not (UpdateGSAction, SaveXMLAction). I see the followign priorities:
- make the XMLConfigWriter.store method synchronized, no two threads should ever enter that method togheter
- externalize the save and update action code into an utility class that can be reused by IE and/or by RESTConfig, make the two calls (update and save)
synchronized so that it's not possible to carry either one in paraller (with itself or with the other one).
Those changes are trivial and should go a long way to ensuring multiple admins do not screw up the configuration.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.codehaus.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira