Hi guys. A bit "edge" topic to this mailing list, but still.
I'd like to run Spark on Mesos in a way that mixes some jobs using the dynamic allocation and others being "static" owners of their resources. I don't see any way Spark offers this natively. The idea is turning off the shuffle service in part of the nodes to give them for "static" tasks. I'm not sure that would work, could someone please advise? Or, if you see any other solution to this in a more general way - do share. Thanks a lot, Gregory -- http://ror6ax.github.io/
