I've been trying out Distributed Builds in 1.3.4 and wanted to get some feedback on some concerns I have.
Say I have 10 project groups with 15 projects each, so 150 projects building on various schedules and build definitions. If I set up five build agents and let this run long enough, eventually all 150 projects will get checked out and built on all five agents. That is potentially a *lot* of disk space compared to what we use now with separate instances. Has any thought been given to keeping disk space usage under control on the agents? Right now you can't delete projects that aren't in use anymore [1]. But even if that worked, it seems like there needs to be some scheduled process that gets rid of legitimate working directories that just haven't been used in a while. The local repository is another thing that tends to fill up. I know there is a purge feature for that, does it work on the distributed agents? What else takes up space that we need to think about in terms of the multiplier effect of distributed build? [1] http://jira.codehaus.org/browse/CONTINUUM-2376 Thanks, -- Wendy
