Hi, I have quite a complex scenario which I'm having difficulty in setting up. I've tried multiple different ways and have found issues in all of them and so I'm looking for some inspiration. I'll try my best to describe the situation and then describe how I thought it should be done. I hope someone has the patience to work through this and make suggestions.
- One project which is stored in a single central Mercurial repository. - The project has several branches which all can be build concurrently. - Each branch should be build in three locations at different times of day (US, UK, ASIA) Each location may need to build each branch on several platforms (Linux, solaris) To build a branch the builds is carried out in two stage. - Database build once per branch per location. - Setup a schema on a shared database which can be used by multiple builds. - Code build. One per branch per location per platform (This is a Makefile based c++ project) So the generic structure I have is: 1. Clone Repo to the location(e.g. US, UK, ASIA) we are building. This is done as a shell script not using the mercurial plugin. This is so that multiple builds will end up using the same changeset per location. Clone is in /local-repos/jenkins/. This then kicks of a Database build for the same location, one for each branch. 2. Database build. Parametrised job, takes in the branch to build. Mercurial Plugin uses /local-repos/jekins (this is to make sure that this and the code build use the same changeset). Passes parameters on to Code Build. 3. Code build is a matrix build with axis of platform labels (Linux and Solaris) and has a parameter of which branch to build. It also uses the mercurial plugin and updates from /local-repos/jekins. I've found that I've had to duplicate these three steps for each location which is a shame. I thought it should be possible to create one Database Build and Code Build matrix job which both had location as an axis. Then have the clone step restrict by location when it kicked off the next step. But I found that the database job wouldn't then kick off the code builds step. I think it is because it needs to do all of the configurations in a matrix job before it is classed as complete. Less of an issue is that you cannot tell that the matrix job has succeeded unless they were all kicked off at the same time. I hope this makes some kind of sense. I need to know if there is some really simple method which I've missed, some plugin which might help me. Have I gone about it in completely the wrong way? Is there some document which I haven't found which will describe exactly what I need? Any advice if you can understand what I've tried to describe would be most welcome. Thanks, Stefan -- You received this message because you are subscribed to the Google Groups "Jenkins Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
