Ok, that blows my mind ... I don't see how to achieve this without either pinning the job to a specific server or by composing a tool that keeps track on which machine currently is building. That is an issue I have been cracking my brain on for a while and I did also just see one quick way. I haven't found a tool yet that has this functionality. Tie the job to a machine. That's the easiest way although not charming ;)
take care On 15 Mrz., 11:14, danny staple <[email protected]> wrote: > Not quite. The job is running and not yet archived any artifacts. I want to > pump a file through the tool needing a url as a build step, or in a script > in a build step. > > We've found a crude way using user content dir and taking build number and > job names into a path there which works for now, although it makes it > harder for the job to run anywhere but master. It still allows for > concurrence on multiple executors. > > typed on my phone expect typos... > On 15 Mar 2012 09:34, "Jan Seidel" <[email protected]> wrote: > > > > > > > > > Hi Danny, > > > let me see if I get it right (bit tired and not woken up > > completely ;) ): > > - Your tool has to be supplied with an URL to collect one or more > > files. > > - It has to access a job result. > > - This job has finished to run. > > - It should always be the latest available file you need to access. > > - There are several workspace present for this job, so the job can > > roam to available resources. > > > If that's what you want: > > - your URL will then be > > http:// > > [yourserver:some_port]/job/[JOB_NAME]/lastSuccessfulBuild/artifact/[path_to_artifact]/[file_name] > > > The bracketed part has to be replaced with your data. > > > Take care > > Jan > > > On 14 Mrz., 15:00, Danny Staple <[email protected]> wrote: > > > We have a scenario here where: > > > * A tool on a service can be started via a URL, and one of its parameters > > > is a URl for it to collect a source file from. > > > * We would like to use this tool in a job, such that we create the source > > > for the tool, then want to start the tool above with a file in our > > > workspace. We'd then be long polling, to get the result back from the > > tool. > > > > The thing is, I can't see if there is a clear way to get a workspace file > > > via HTTP if the job is concurrent (which means more than one workspace). > > > Has anyone on the list any ideas for this? Assume for now that the > > service > > > cannot be changed, and that disabling the job concurrency would be > > > prohibitively difficult (we'd have create many separate jobs to get the > > > same functionality without it). > > > > Thanks, > > > Danny
