Add the ability to set the number of different root types indexed in parallel

Description

There are several reasons for which indexing all root types in parallel turns out to be a bad idea.

  • number of needed jdbc connections is unpredictable

  • some other operations could run in parallel (index optimisation for eg)

  • generates a very high amount of contention without a clear benefit

Discussed on http://lists.jboss.org/pipermail/hibernate-dev/2013-August/010229.html and related to

Environment

None

Activity

Show:
Guillaume Smet
August 6, 2013, 1:25 PM

Yes, it's a local PostgreSQL database. I'm running it with your suggested configuration and I can confirm that the CPU are used at 100%, which is what I had with 3 indexing threads and my original configuration.

Are we OK with the API, especially the name of the fluid setter? If so, I'm planning to update the paragraph "6.3.2. Using a MassIndexer" of the documentation.

Guillaume Smet
August 6, 2013, 3:50 PM

Sanne, with your settings, indexing took 10 minutes. Looks like I should change my default configuration .

Sanne Grinovero
August 6, 2013, 4:08 PM

won the bet

Guillaume Smet
August 12, 2013, 12:59 PM

Sanne,

I changed the method name and updated the documentation. If everything is OK for you, I can rebase and rewrite the history of my branch to obtain a clean commit before you merge.

Thanks for your feedback.


Guillaume

Sanne Grinovero
August 12, 2013, 2:17 PM

Thanks! looks like perfect, merging.

Assignee

Guillaume Smet

Reporter

Sanne Grinovero

Labels

None

Suitable for new contributors

None

Pull Request

None

Feedback Requested

None

Components

Fix versions

Priority

Major
Configure