-
Bug
-
Resolution: Won't Fix
-
Major
-
None
-
Centos 7.1.1503, Jenkins 1.620, Docker plugin 0.10.0, Docker 1.7.1
We have 2 Docker hosts configured as Docker Clouds in our master Jenkins box, mainly for load balancing and failover of jobs. We are using "Keep this slave on-line as much as possible" availability setting. Both hosts are set up with exactly the same set of Docker images/containers/labels and are restricted to run a single instance of each container.
To start from scratch and for testing purposes, I deleted all slaves from the Jenkins master. Next, when I launch number of jobs restricted to run on particular label (which is available on both hosts), the Docker container for the corresponding label only starts on one of the hosts and the jobs are then queued up and run in serial on this single slave.
The expected behaviour would be to spin up as much as possible Docker containers (within set limits) and run the queued jobs in parallel on multiple slaves.
Please let me know if you need any further information from me.
Well this retention strategy is experimental, but your issue may be related to NodeProvisioner logic.
I have no ideas atm.