jdef

at this point in time, k8s has beta support for autoscaling that requires the heapster add-on to be deployed: http://kubernetes.io/v1.1/docs/user-guide/horizontal-pod-autoscaler.html TODO: enable heapster in our mesos/docker cluster and start running...

#424

the scheduler needs to expose a consistent artifact port across all instances. so using port 0 should not be allowed AT ALL when HA/failover is configured.

reproduced on latest upstream_k8sm build. in this case it appears that the slave recognizes that it was asked to kill the task, but the executor never logged that it received...

``` shell $ dpkg -l |grep -e mesos ii mesos 0.21.1-1.1.ubuntu1404 amd64 Cluster resource manager with efficient resource isolation $ uname -a Linux node-1 3.13.0-29-generic #53-Ubuntu SMP Wed Jun 4...

xref https://issues.apache.org/jira/browse/MESOS-2865

and reconciliation isn't cleaning up the mess either (probably because the tasks have been flagged as `Deleted` so perhaps the reconciliation mechanism assumes that a kill is in progress): ```...

mesos-slave.WARN within 2m of the "task kill" event: ``` W0612 03:03:33.358157 1328 status_update_manager.cpp:472] Resending status update TASK_LOST (UUID: 86e13b65-1034-417a-86d2-f5c323e42c56) for task pod.a8ba5838-0471-11e5-96e5-525400309a8f of framework 20150511-114 826-83886602-5050-28892-0001 W0612 03:03:33.358258 1328 status_update_manager.cpp:472]...

actually, the minion/server.go:launchExecutorServer() code never tries to restart the executor. once it dies, the minion tears down kube-proxy and then exits. On Sat, Oct 24, 2015 at 2:05 AM, Dr....

xref #410 (duplicate?)