k8sm launch task failed

[yehuis4]

nginx_pod.yml:

apiVersion: v1
kind: Pod
metadata:
  name: nginx
spec:
  containers:
    - name: nginx
      image: nginx
  ports:
    - containerPort: 80
      hostPort: 31320
      resources:
        limits:
        cpu: 750m
        memory: 1Gi

Steps and error details:

Step1. mesos-slave specified --default_role=someone. we could see the following logs of mesos-master，mesos-slave declared all the resources with role of someone:

I0914 11:47:03.723081 1315 master.cpp:3000] Registered slave 20150914-111917-302055713-5050-1294-S0 at slave(1)@33.1.1.19:5051 (cluster-4) with cpus(someone):2; mem
(someone):2929; disk(someone):35164; ports(someone):[31000-32000]
I0914 11:47:03.723228 1315 hierarchical.hpp:455] Added slave 20150914-111917-302055713-5050-1294-S0 (cluster-4) with cpus(someone):2; mem(someone):2929; disk(someon
e):35164; ports(someone):[31000-32000] (and cpus(someone):2; mem(someone):2929; disk(someone):35164; ports(someone):[31000-32000] available)

Step2. and then the master sent offers to k8s， also with “someone”. The log‘s as follows：

I0914 11:52:16.592625 1310 master.cpp:3760] Sending 1 offers to framework 20150914-111917-302055713-5050-1294-0000 (Kubernetes) at scheduler(1)@33.1.1.22:55004
I0914 11:52:21.003809 1315 master.cpp:3461] Performing implicit task state reconciliation for framework 20150914-111917-302055713-5050-1294-0000 (Kubernetes) at sch
eduler(1)@33.1.1.22:55004
I0914 11:52:21.596210 1312 master.cpp:2273] Processing ACCEPT call for offers: [ 20150914-111917-302055713-5050-1294-O1 ] on slave 20150914-111917-302055713-5050-12
94-S0 at slave(1)@33.1.1.19:5051 (cluster-4) for framework 20150914-111917-302055713-5050-1294-0000 (Kubernetes) at scheduler(1)@33.1.1.22:55004
I0914 11:52:21.597126 1312 hierarchical.hpp:648] Recovered cpus(someone):2; mem(someone):2929; disk(someone):35164; ports(someone):[31000-32000] (total allocatable:
cpus(someone):2; mem(someone):2929; disk(someone):35164; ports(someone):[31000-32000]) on slave 20150914-111917-302055713-5050-1294-S0 from framework 20150914-11191
7-302055713-5050-1294-0000

Step3. k8sm create task(e.g. create pod). And the task failed with below log(mesos-master.log):

I0915 12:02:19.634052 1309 master.cpp:3336] Sending status update TASK_ERROR (UUID: 3fa1333c-fdfd-49dc-baf8-b1a989cd3a87) for task pod.9d458b74-5ba1-11e5-9b03-fa163
edb51bd of framework 20150914-111917-302055713-5050-1294-0000 'Task uses more resources cpus(*):1; mem(*):1088; ports(*):[31320-31320] than available cpus(someone):2; mem(someone):2929; disk(someone):35164; ports(someone):[31000-32000]'

Am I did something in wrong way? Or is it a bug?
If it is a bug, can I fix it and submmit a PR?

[j3ffml]

cc @kubernetes/kube-mesos

[jdef]

known bug: mesosphere/kubernetes-mesos#482

[yehuis4]

Thanks a lot. We expect that you can fix this bug ASAP.

[jdef]

It's on our roadmap. We also accept PR's :)

On Tue, Sep 15, 2015 at 8:57 PM, yehui notifications@github.com wrote:

Thanks a lot. We expect that you can solve this problem ASAP.

—
Reply to this email directly or view it on GitHub
#13956 (comment)
.

[jdef]

closing this out since it's a duplicate of the aforementioned bug we're already tracking on our roadmap