On Mon, Jan 18, 2016 at 11:39:23AM +0000, John Williams wrote: > > I have noticed a few times, including as I write this, that sometimes the > progress of a build seems to get stuck as shown at: > > http://ci.netsurf-browser.org/jenkins/view/All/job/netsurf/changes > > Today the progress is still stuck from Jan 15, 2016 7:37 on #3269 at, say, > 95% completion, and a build #3270 remains "pending". > > RISC OS #3269 has downloaded fine, but presumably something has stuck at a > later point for another OS. > > Why/how does this happen?
The CI system is built around the concepts of jobs, each job represents a task the system needs to perform such as building NetSurf binaries or packages etc. Each job requires some resources usually provided by one or more worker systems. So for example the master branch NetSurf build requires a worker (we are transitioning to using the controller/worker naming from master/slave after some very spirited complaints) for each compiled architecture (e.g. cislave4 compiles for arm-unknown-riscos). What you are seeing in your example is where a resources is no longer unavailable to the CI system (specifically cislave3 running the haiku OS). The failure means the in progress job cannot complete and new jobs that require that resource cannot commence. It requires an administrator (me) to come along and reboot the failed system and restart the CI worker software to make the resources available again and allow the job to complete. > > Also, sometimes there is nothing listed under Changes for versions - not > even a build reference. > Builds can be manually started by developers if there has been an issue during a build, these will have no code changes and be anonymous > Just curious! > > John > > -- > | John Williams > | joh...@ukgateway.net > > I think, therefore I am unsure - I think! * > > -- Regards Vincent http://www.kyllikki.org/