The scheduler currently requires/assumes that two fault domains are identified on slaves: host and rack. This is done by supplying the 'attributes' argument to mesos-slave. You can see an example of this in our vagrant script for the slave [1]. Hope that helps!
[1] https://github.com/apache/incubator-aurora/blob/master/examples/vagrant/upstart/mesos-slave.conf -=Bill On Fri, Sep 12, 2014 at 8:33 AM, Olivier Sallou <olivier.sal...@irisa.fr> wrote: > Hi, > this is my first aurora install but I gonna get it work. > I have installed mesos and aurora on a single server (for test). > > Mesos is installed and running fine (I tested a sample job). > The Web UI is fine too and when I submit a job (hello_world from doc) I > can see it in the UI, but my job remains in pending state. > > PENDING : Constraint not satisfied: host > > Job configuration is: > > configuration details for instances 0 > resources cpu 1 cores > ram 1.00 MiB > disk 8.00 MiB > constraints host:{"limit":1} > production false > service true > > My compuer is multi core (mostly idle), so I do not think it is a > resource issue.... > > Any help would be welcomed... > > Thanks > > Olivier > > -- > > > gpg key id: 4096R/326D8438 (keyring.debian.org) > Key fingerprint = 5FB4 6F83 D3B9 5204 6335 D26D 78DC 68DB 326D 8438 > >
