github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-07-23 19:56:01 +00:00
Code Issues Projects Releases Wiki Activity

Merge pull request #51282 from shyamjvs/new-allowed-not-ready-semantics

Browse Source 
Automatic merge from submit-queue

AllowedNotReadyNodes allowed to be not ready for absolutely *any* reason

It's as good as we allow those many nodes to be not part of the cluster at all, ever.

Btw - currently our 5k-node correctness test fails if "kubelet stopped posting node status" or "route not created", etc (ref: https://storage.googleapis.com/kubernetes-jenkins/logs/ci-kubernetes-e2e-gce-scale-correctness/3/build-log.txt)

cc @kubernetes/sig-scalability-misc
This commit is contained in:
Kubernetes Submit Queue 2017-08-25 05:00:32 -07:00 committed by GitHub
commit 11299e363c
2 changed files with 3 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
test/e2e/framework/BUILD
View File

@ -43,7 +43,6 @@ go_library(
"//pkg/api:go_default_library",
"//pkg/api/testapi:go_default_library",
"//pkg/api/v1/helper:go_default_library",
"//pkg/api/v1/node:go_default_library",
"//pkg/api/v1/pod:go_default_library",
"//pkg/apis/batch:go_default_library",
"//pkg/apis/extensions:go_default_library",

32
test/e2e/framework/util.go
View File

@ -75,7 +75,6 @@ import (
clientset "k8s.io/client-go/kubernetes"
"k8s.io/kubernetes/pkg/api"
"k8s.io/kubernetes/pkg/api/testapi"
nodeutil "k8s.io/kubernetes/pkg/api/v1/node"
podutil "k8s.io/kubernetes/pkg/api/v1/pod"
batchinternal "k8s.io/kubernetes/pkg/apis/batch"
extensionsinternal "k8s.io/kubernetes/pkg/apis/extensions"
@ -2487,10 +2486,7 @@ func WaitForAllNodesSchedulable(c clientset.Interface, timeout time.Duration) er
Logf("================================")
}
}
if len(notSchedulable) > TestContext.AllowedNotReadyNodes {
return false, nil
}
return allowedNotReadyReasons(notSchedulable), nil
return len(notSchedulable) <= TestContext.AllowedNotReadyNodes, nil
})
}
@ -3597,23 +3593,6 @@ func WaitForNodeToBe(c clientset.Interface, name string, conditionType v1.NodeCo
return false
}
// Checks whether not-ready nodes can be ignored while checking if all nodes are
// ready (we allow e.g. for incorrect provisioning of some small percentage of nodes
// while validating cluster, and those nodes may never become healthy).
// Currently we allow only for:
// - not present CNI plugins on node
// TODO: we should extend it for other reasons.
func allowedNotReadyReasons(nodes []*v1.Node) bool {
for _, node := range nodes {
index, condition := nodeutil.GetNodeCondition(&node.Status, v1.NodeReady)
if index == -1 ||
!strings.Contains(condition.Message, "could not locate kubenet required CNI plugins") {
return false
}
}
return true
}
// Checks whether all registered nodes are ready.
// TODO: we should change the AllNodesReady call in AfterEach to WaitForAllNodesHealthy,
// and figure out how to do it in a configurable way, as we can't expect all setups to run
@ -3643,19 +3622,14 @@ func AllNodesReady(c clientset.Interface, timeout time.Duration) error {
// of nodes (which we allow in cluster validation). Some nodes that are not
// provisioned correctly at startup will never become ready (e.g. when something
// won't install correctly), so we can't expect them to be ready at any point.
//
// However, we only allow non-ready nodes with some specific reasons.
if len(notReady) > TestContext.AllowedNotReadyNodes {
return false, nil
}
return allowedNotReadyReasons(notReady), nil
return len(notReady) <= TestContext.AllowedNotReadyNodes, nil
})
if err != nil && err != wait.ErrWaitTimeout {
return err
}
if len(notReady) > TestContext.AllowedNotReadyNodes || !allowedNotReadyReasons(notReady) {
if len(notReady) > TestContext.AllowedNotReadyNodes {
msg := ""
for _, node := range notReady {
msg = fmt.Sprintf("%s, %s", msg, node.Name)