Hello,
I guess it's okay to create a new group in Hudson for Derby?
I'm trying out the Hudson plugin in NetBeans, and if there is no group
for the builds you're interested in it seems you have to keep an eye on
all builds in the instance. That's a waste of resources (assuming it's
smart enough to only query for the groups being shown).
The Derby project has a test suite that takes between 2 - 4 hours to run
(2.8 GHz dual core/CPU), and at times it can be a bit CPU and/or IO
intensive. Most tests are single threaded, but there are a few that
would keep more than one CPU busy as well. Run on a low-frequency
CMT-machine the suite would take longer, and the toll on the IO
subsystem would probably be higher than on the CPU.
I chose not to run this in Hudson, as it would block the other projects
for a long time. The Derby jobs are currently tied to Solaris, but they
can be run on any platform.
My questions:
- is there enough resources to run the test suite in Hudson, say once a
week?
(this would allow us to get test coverage reports as well)
- would it help if Derby requested a zone and "donated" (as Lucene has
done) it to Hudson, or is the zones machine already saturated?
The Derby project is currently reasonably well covered outside of Apache
infrastructure when it comes to testing, but that process is a lot more
closed than it would be if run on Apache infrastructure. Compiling Derby
and building the documentation in Hudson were the two first steps
towards a more transparent process / community. I just want to provide a
minimal level of services on ASF infrastructure, we'll still be
dependent on external resources to cover our needs.
Regards,
--
Kristian
