[k8s.io] ConfigMap should be consumable via environment variable [Conformance] {E2eNode Suite}

[k8s-github-robot]

https://k8s-gubernator.appspot.com/build/kubernetes-jenkins/logs/ci-kubernetes-node-kubelet/2498/
Failed: [k8s.io] ConfigMap should be consumable via environment variable [Conformance] {E2eNode Suite}

/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/common/configmap.go:382
Expected error:
    <*errors.errorString | 0xc4202685a0>: {
        s: "failed to get logs from pod-configmaps-59ebd4af-087b-11e7-90c9-42010a800019 for env-test: the server could not find the requested resource (get pods pod-configmaps-59ebd4af-087b-11e7-90c9-42010a800019)",
    }
    failed to get logs from pod-configmaps-59ebd4af-087b-11e7-90c9-42010a800019 for env-test: the server could not find the requested resource (get pods pod-configmaps-59ebd4af-087b-11e7-90c9-42010a800019)
not to have occurred
/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/framework/util.go:2195

Previous issues for this test: #42526

[ncdc]

This appears to have found a segfault:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x20 pc=0xddd7fb]
goroutine 71645 [running]:
panic(0x2d90800, 0xc42000e040)
        /usr/local/go/src/runtime/panic.go:500 +0x1a1
k8s.io/kubernetes/pkg/kubelet/cm.(*cgroupManagerImpl).Pids.func1(0xc420d33860, 0x9d, 0x0, 0x0, 0x4d247e0, 0xc421d0aa50, 0x0, 0x0)
        /go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/pkg/kubelet/cm/cgroup_manager_linux.go:452 +0x3b
path/filepath.walk(0xc420d334a0, 0x97, 0x4d48b00, 0xc421ba5e10, 0xc421157880, 0x0, 0x0)
        /usr/local/go/src/path/filepath/path.go:372 +0x22e
path/filepath.walk(0xc4221f73e0, 0x56, 0x4d48b00, 0xc421ba5d40, 0xc421157880, 0x0, 0x1)
        /usr/local/go/src/path/filepath/path.go:376 +0x344
path/filepath.Walk(0xc4221f73e0, 0x56, 0xc421157880, 0x0, 0x0)
        /usr/local/go/src/path/filepath/path.go:398 +0xd5
k8s.io/kubernetes/pkg/kubelet/cm.(*cgroupManagerImpl).Pids(0xc42038ab20, 0xc421b19440, 0x3c, 0xc42108b728, 0xc4223dafc0, 0x1)
        /go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/pkg/kubelet/cm/cgroup_manager_linux.go:466 +0x380
k8s.io/kubernetes/pkg/kubelet/cm.(*podContainerManagerImpl).tryKillingCgroupProcesses(0xc421f08960, 0xc421b19440, 0x3c, 0x43d515, 0xc42108b6d8)
        /go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/pkg/kubelet/cm/pod_container_manager_linux.go:120 +0x7e
k8s.io/kubernetes/pkg/kubelet/cm.(*podContainerManagerImpl).Destroy(0xc421f08960, 0xc421b19440, 0x3c, 0x1, 0xc421fad540)
        /go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/pkg/kubelet/cm/pod_container_manager_linux.go:157 +0x5a
created by k8s.io/kubernetes/pkg/kubelet.(*Kubelet).cleanupOrphanedPodCgroups
        /go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/pkg/kubelet/kubelet_pods.go:1554 +0x381

cc @kubernetes/sig-node-bugs @derekwaynecarr @sjenning @dchen1107 @yujuhong @vishh

[ncdc]

If I had to guess, in between the call to os.IsNotExist at

kubernetes/pkg/kubelet/cm/cgroup_manager_linux.go

Line 438 in ccd87fc

if os.IsNotExist(err) {

and the call to filepath.Walk at

kubernetes/pkg/kubelet/cm/cgroup_manager_linux.go

Line 466 in ccd87fc

if err = filepath.Walk(dir, visitor); err != nil {

, the dir in question is deleted, which results in a nil info at

kubernetes/pkg/kubelet/cm/cgroup_manager_linux.go

Line 452 in ccd87fc

if !info.IsDir() {

.

[ncdc]

Raising priority because this can cause the kubelet to crash

[sttts]

Confirming @ncdc's analysis. If any stat fails, the WalkFn is called with the err and info=nil.

[ncdc]

I will post a PR shortly

[ncdc]

I'm going to open a new issue for the root cause, as this one isn't descriptive for it

[ncdc]

Reassigning as @Random-Liu has the fix PR