Migration Tools

OpenOffice.org migration will most likely be based on using a modified and for OpenOffice.org enhanced and adapted version of the netbeans migration tools also know as netbeans migration scripts which have been used when netbeans migrated from a CEE/collabnet based website to a kenai based website.

The netbeans migration tools consist of some unix shell scripts and programs in the java, ruby and scala programming languages and SQL procedure which are used for migration and testing. Additionally the netbeans migration tools contain the netbeans specific configuration and customization files for bugzilla.

Besides programs that are responsible for the actual migration the netbeans migration tools also contain test programs for checking the result after migration and stress testing.

We will get a dump of data for the current OpenOffice.org site from collabnet, this includes a dump of the 'tigris' mysql database, which includes things like project definitions, users, roles and mailing lists, and a dump of the issuetracker database. Data about subscribed users to mailing lists will most likely not be included in the data dump from collabnet. The netbeans migration tools contain a script to load the information about subscribed users via the web admin frontend of a CEE based website.

Most of the netbeans migration programs do migration of data in two steps: first data is 'unloaded' from a database which was loaded with the database dumps from Collabnet or from the administration pages of the old CEE website into yaml files and than this data is 'loaded' via ruby scripts into the new kenai instance.

The ruby scripts of the migration tools must be executed in a kenai junction2 context and that means they need to be copied to kenai~app/junction2/db/scripts before they can be used.

The netbeans migration tools do also contain a program to incrementally update the local copy of the tigris database that was created from the collabnet dump to a current state by parsing HTML files which contain current data created by using the AdHocQuery feature available to domain administrators on a CEE website.

Also included in the netbeans migration tools is a an apache module they used for customizing netbeans Wiki authentication which we might also later make use of to implement a single sign on for our OpenOffice.org wiki which is on a satellite website and will not be migrated to a kenai wiki. Netbeans has a similar setup with their own MediaWiki deployment instead of using the Kenai provided Wiki.

Also included in the netbeans migration tools is a script to check the URLs of the website content.

Some topics for migration OOo has are not covered by the netbeans migration tools such as migrating "canconfirm" permissions for issuezilla for these new programs must be created by the OOo migration team.

In the netbeans migration tool there are also some very netbeans specific things like deleting certain users and also there are also things like hardcoded paths or database connection settings. There are also some differences between the CEE tigris database and the OpenOffice.org tigris database. As seen so far OpenOffice.org seems to some additional columns in some of the tables. All in all this means that the netbeans programs can not be used unmodified but must be evaluated and adapted carefully to OpenOffice.org needs. For doing this we decided not to make the extra effort of creating a more generalized version of the netbeans migration tools but to create just our own OpenOffice.org migration tools but to reuse as much as possible from the netbeans tools.

Open Questions

• do we also need incremental update to our copy of the OpenOffice.org tigris database?
• the netbeans team seems to have manually used the AdHocQuery feature and than saved the data for later parsing by programs in the netbeans migration tools for incremental update. Do we also want to automate that step and use a shell script with some curl calls for example to call the AdHocQuery form and download the result HTML files?
• ...