Exactly. Copying artifacts to master for archiving and copying them out to the workspace where they are used seems to be the way Jenkins works, at least, currently. You need to install the Copy artifact plugin, or at least it will make your life easier.
At work my build artifacts are around 300 MB and I haven't considered the time spent copying them to be a show stopper. I get an archived copy of the artifacts and I do not have worry about something overwriting the artifacts. If your artifacts really are so large they cannot be moved around, I believe you need to create a new plugin to make it safe to pass them from one job to another in one slave. -- Sami Les Mikesell <[email protected]> kirjoitti 19.6.2012 kello 23.48: > On Tue, Jun 19, 2012 at 2:57 PM, Jeff <[email protected]> wrote: >> I have a potential use case....but I'm also new to Jenkins so this may be >> doable today in a way that I have yet to learn. >> >> In trying to resolve how to use Maven and Jenkins and create a continuous >> delivery (CD) pipeline where each step in the pipeline progresses the state >> such as: build->Unit Test->Deploy-dev->Integration Tests->Deploy-stage->Load >> Tests->DeployLive (or something like this), we could have the functionality >> to run each pipeline stage as part of the same source project and configured >> via various profiles/plugins in the Maven POM. >> >> However, it seems that allowing any random slave to execute the various >> steps in the pipeline would incur a lot of unneeded overhead by either >> requiring the project and artifacts be archived and/or copied to each slave >> at each point in the pipeline depending on who gets the job, or the same >> source would need to be checked out and built for whichever phase is being >> run. Seems like a lot of unneeded overhead. > > The overhead of a few file copies is pretty small in the context of > builds and probably more than offset but letting you build components > in parallel and especially if you can come up with a scheme to only > build the changed component(s) and the final result. > > -- > Les Mikesell > [email protected]
