- refactor queue.Pod construction to take functional options, privatize Pod fields
- refactor DelayFIFO and HistoricalFIFO to offer consistent, more useful Pop() funcs
- refactor pod update processing changes; long term we should somehow combine with the special pod config source that we are using for mirror pods
- task launch timer cleanup
- add busybox static pod to mesos-docker cluster
- customize static pods with binding annotations
- code cleanup
- removed hacky podtask.And func; support minimal resources for static pods when resource accounting is disabled
- removed zip archive of static pods, changed to gzip of PodList json
- pod utilities moved to package podutil
- added e2e test
- merge watched mirror pods into the mesos pod config stream
This avoid that we either waste cpu resources due to rounding or that we report
to much to the kubelet such that the e2e tests think they can schedule more than
resources are available.
This fixes https://github.com/mesosphere/kubernetes-mesos/issues/437
This patch reduces the dependencies of the executor from the kubelet. This
makes it possible launch the kubelet after the executor.
This considerably reduces the complexity of the startup code.
Moreover, this work is a requirement to use a standalone kubelet some day.
- pre-create node api objects from the scheduler when offers arrive
- decline offers until nodes a registered
- turn slave attributes as k8s.mesosphere.io/attribute-* labels
- update labels from executor Register/Reregister
- watch nodes in scheduler to make non-Mesos labels available for NodeSelector matching
- add unit tests for label predicate
- add e2e test to check that slave attributes really end up as node labels
- instantiate framework.Controller for pods in the executor using framework.NewInformer,
in order to watch pod updates for pods on that host
- forwards updates like graceful termination to the kubelet.
This might also be the preparation for other updates which are supported by the
kubelet.
Before NodeName in the pod spec was used. Hence, pods with a fixed, pre-set
NodeName were never scheduled by the k8sm-scheduler, leading e.g. to a failing
e2e intra-pod test.
Fixesmesosphere/kubernetes-mesos#388
- the mesos scheduler gets a --static-pods-config parameter with a directory with
pods specs. They are zipped and sent over to newly started mesos executors.
- the mesos executor receives the zipper static pod config via ExecutorInfo.Data
and starts up the pods via the kubelet FileSource mechanism.
- both - the scheduler and the executor side - are fully unit tested
