odd behavior on Centos7 box

[Brendan Smith]

I mentioned having issues on CentOS7 in another thread and was told to open a new ticket, so here it is... i'm not entirely sure it's CentOS7 related or not, but I figured I'd throw it out there and see if anyone else is having issues.

for reference, I am running puppet-mesos v0.8.0, mesos v0.27.1 and provisioning with Puppet 4

my biggest road block right now is the way mesos options are setup on the master.. If I provision this box with current version of puppet-mesos, my configuration options end up looking like:

ls /etc/mesos-master/
master_acls  master_hostname  master_log_dir  master_max_completed_tasks_per_framework  master_quorum  work_dir

oddly, work_dir gets passed in normally, even though it's defined as master_work_dir in master.pp

Which then get passed to the mesos init script as --master_acls --master_hostname ... which are obviously the wrong flags. If i remove this 'master' the options get passed in as expected, ie:

ls /etc/mesos-master/
acls  hostname  log_dir  max_completed_tasks_per_framework  quorum  work_dir

these then get passed in as normal to the init script, but i run into other issues which i will continue to debug.

Does this sound like a CentOS7 specific issue?

[Brendan Smith]

looking at some code more now i see why work_dir is working as expected.

@deric do you know if the Ubuntu package expects these options to be 'master_' / 'slave_' and the CentOS package does not? Or is this broken across the board?