I'm thinking out loud here, but I wonder if it would be nice to be able to support the following scenarios in Jenkins (and docker plugin users have been asking for similar features).
I use docker to launch slaves on a variety of boxes. This works very nicely. You can tell docker to attach a 'data only' container (e.g: for the workspace), but you can only do this at the point where you launch the container. What would be nice is to launch a slave with a blank 'workspace' container. Once that build is complete, the workspace container is tagged and stored. The next stage of the pipeline is invoked, and the slave is launched *with that tagged workspace attached*. This would be nice as it would avoid a lot of the artifact copying that goes on -- and would allow users to split up multiple maven invocations across build steps. The sticking bit seems to be the requirement to know - at the time the slave is launched - which workspace name to attach. The current API (unless I'm missing something) seems to be more around "this job demands any slave (possibly with an optional label). If demand > supply, try to launch more slaves". Effectively, every build may need a 'custom kind' of slave. Secondly it may be useful to be able to say "only launch downstream slaves on the same physical node as the upstream job". Thoughts? -- You received this message because you are subscribed to the Google Groups "Jenkins Developers" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
