github/kubernetes
4
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Use node out of disk condition in the scheduler while scheduling pods.

Browse Source 
Set the out of disk node condition to unknown in the node controller if
the kubelet does not report its node condition in a long time. Update
node controller unit tests.

Implement a node condition predicate function that checks if a given
node satisfies the conditions defined by the predicate and if it
does, use that node for scheduling pods. The predicate function takes
both NodeReady and NodeOutOfDisk into consideration to determine if a
node is fit for scheduling pods.

The predicate is then passed to the node lister in the scheduler factory
so that the node lister can run the predicate function on the nodes when
schedling pods thereby omitting nodes that does not satisfy the
predicate.

Also update listers test.
This commit is contained in:
Madhusudan.C.S
2015-10-22 12:47:43 -07:00
parent b74b2aa43e
commit ce257b5a0e
5 changed files with 171 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
pkg/controller/node/nodecontroller.go
View File

@@ -614,7 +614,7 @@ func (nc *NodeController) tryUpdateNodeStatus(node *api.Node) (time.Duration, ap
if nc.now().After(savedNodeStatus.probeTimestamp.Add(gracePeriod)) {
// NodeReady condition was last set longer ago than gracePeriod, so update it to Unknown
// (regardless of its current value) in the master, without contacting kubelet.
// (regardless of its current value) in the master.
if readyCondition == nil {
glog.V(2).Infof("node %v is never updated by kubelet", node.Name)
node.Status.Conditions = append(node.Status.Conditions, api.NodeCondition{
@@ -637,6 +637,32 @@ func (nc *NodeController) tryUpdateNodeStatus(node *api.Node) (time.Duration, ap
readyCondition.LastTransitionTime = nc.now()
}
}
// Like NodeReady condition, NodeOutOfDisk was last set longer ago than gracePeriod, so update
// it to Unknown (regardless of its current value) in the master.
// TODO(madhusudancs): Refactor this with readyCondition to remove duplicated code.
oodCondition := nc.getCondition(&node.Status, api.NodeOutOfDisk)
if oodCondition == nil {
glog.V(2).Infof("Out of disk condition of node %v is never updated by kubelet", node.Name)
node.Status.Conditions = append(node.Status.Conditions, api.NodeCondition{
Type: api.NodeOutOfDisk,
Status: api.ConditionUnknown,
Reason: "NodeStatusNeverUpdated",
Message: fmt.Sprintf("Kubelet never posted node status."),
LastHeartbeatTime: node.CreationTimestamp,
LastTransitionTime: nc.now(),
})
} else {
glog.V(2).Infof("node %v hasn't been updated for %+v. Last out of disk condition is: %+v",
node.Name, nc.now().Time.Sub(savedNodeStatus.probeTimestamp.Time), oodCondition)
if oodCondition.Status != api.ConditionUnknown {
oodCondition.Status = api.ConditionUnknown
oodCondition.Reason = "NodeStatusUnknown"
oodCondition.Message = fmt.Sprintf("Kubelet stopped posting node status.")
oodCondition.LastTransitionTime = nc.now()
}
}
if !api.Semantic.DeepEqual(nc.getCondition(&node.Status, api.NodeReady), lastReadyCondition) {
if _, err = nc.kubeClient.Nodes().UpdateStatus(node); err != nil {
glog.Errorf("Error updating node %s: %v", node.Name, err)