OK, I found this is a known issue, https://issues.apache.org/jira/browse/MESOS-435
For me, the problem caused by below line in /etc/hosts 127.0.1.1 vm-mesos-node1 And I have DDNS name setup for vm-mesos-node1(the master) and vm-mesos-node2(the slave), so I always use their DNS name to configure the cluster rather than their IP address. However, since it always lookup IP for hostname in /etc/hosts, it will get 127.0.1.1 for itself, rather than the real IP. To workaround, just remove that line in /etc/hosts works for me. -- Thanks, Chengwei On Fri, Mar 21, 2014 at 09:55:22AM +0800, Chengwei Yang wrote: > Hi, > > I have setup a fresh mesos cluster with one master and one slave, and > I found below log in my mesos master log, it repeat continuously > > ... repeat below, once per second ... > > I0321 09:47:44.818825 6953 master.cpp:2895] Removing slave > 201403210942-16842879-5050-6944-296 (vm-mesos-node2) > I0321 09:47:44.819260 6953 master.cpp:1812] Attempting to register > slave on vm-mesos-node2 at slave(1)@127.0.1.1:5051 > I0321 09:47:44.819592 6953 master.cpp:2782] Adding slave > 201403210942-16842879-5050-6944-297 at vm-mesos-node2 with cpus(*):1; > mem(*):994; disk(*):8699; ports(*):[31000-32000] > I0321 09:47:44.820227 6953 hierarchical_allocator_process.hpp:445] > Added slave 201403210942-16842879-5050-6944-297 (vm-mesos-node2) with > cpus(*):1; mem(*):994; disk(*):8699; ports(*):[31000-32000] (and > cpus(*):1; mem(*):994; disk(*):8699; ports(*):[31000-32000] available) > I0321 09:47:44.820806 6949 master.cpp:619] Slave > 201403210942-16842879-5050-6944-297 (vm-mesos-node2) disconnected > I0321 09:47:44.821141 6949 hierarchical_allocator_process.hpp:484] > Slave 201403210942-16842879-5050-6944-297 disconnected > > It seems the slave is keeping been removed and then registered, is this > behaviour correct? > > -- > Thanks, > Chengwei
