There are plugins to help you do that. Mainly there is the matrix job plugin which allows you to run the same job accordin to different axis (the axis you are looking for is node or label)
By default the queueing system of jenkins tend to reuse the same node it executed the job before to prevent the workspace from being recopied over and over, hense saving disk space. What you want though is the same job run on multiple environment, and mateix build/job should be able to help you in this kind of setup. The name of the job should really no impact on the job distribution to node. you might just fell in a case where the node executors were linearily executing one after the other once they were available. Does that help you ? On 2012-10-05, at 17:21, zperry <zack.pe...@sbcglobal.net> wrote: > We are POCing a Jenkins build farm that consists of 1 master, 2 slaves. > > We use Jenkins mainly to monitor our C++ builds. Our software targets many > different Linux and FreeBSD distributions and releases, so we can see that > Jenkins can help boost our efficiency significantly. > > But while testing it out, it seems to me that even for the same build job, I > have name jobs differently. I wonder it's because I have misunderstood > Jenkins' usage or it's a current limitation that I have to live with? > > To make it concrete, lets say we have > m0, s0, and s1, where m0 is the master, s0 and s1 are the two slaves. > j0, j1, ... jn as the respective "original" name of our C++ build/test jobs > (i.e. when these jobs were ran just on the POC master) > Now, to use s0 and s1, it seems that we must rename our jobs to the following: > On m0: m0-j0, m0-j1, ... m0-jn > On s0: s0-j0, s0-j1, ... s0-jn > On s1: s1-j0, s1-j1, ... s1-jn > This seems to be quite cumbersome an requirement. Can we do it in a more > Object-Oriented manner - use an attribute somewhere to denote a j0 should be > run on s0, and another j0 should be run on s1 etc, likewise for other jobs? > > Thanks in advance for any hints/tips. > > Regards, > > -- Zack >
