github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-08-25 19:43:22 +00:00
Code Issues Projects Releases Wiki Activity

handle task lost status update from executor on docker daemon crash/restart by deleting the pod, so that the RC creates a new one, if the pod is backed by RC

Browse Source
This commit is contained in:
Ravi L R 2015-10-14 23:14:22 -07:00
parent 0e296c3dcd
commit 99b03baa9b
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
contrib/mesos/pkg/scheduler/scheduler.go
View File

@ -437,11 +437,13 @@ func (k *KubernetesScheduler) reconcileTerminalTask(driver bindings.SchedulerDri
if (state == podtask.StateRunning || state == podtask.StatePending) && if (state == podtask.StateRunning || state == podtask.StatePending) &&
((taskStatus.GetSource() == mesos.TaskStatus_SOURCE_MASTER && taskStatus.GetReason() == mesos.TaskStatus_REASON_RECONCILIATION) || ((taskStatus.GetSource() == mesos.TaskStatus_SOURCE_MASTER && taskStatus.GetReason() == mesos.TaskStatus_REASON_RECONCILIATION) ||
(taskStatus.GetSource() == mesos.TaskStatus_SOURCE_SLAVE && taskStatus.GetReason() == mesos.TaskStatus_REASON_EXECUTOR_TERMINATED) || (taskStatus.GetSource() == mesos.TaskStatus_SOURCE_SLAVE && taskStatus.GetReason() == mesos.TaskStatus_REASON_EXECUTOR_TERMINATED) ||
(taskStatus.GetSource() == mesos.TaskStatus_SOURCE_SLAVE && taskStatus.GetReason() == mesos.TaskStatus_REASON_EXECUTOR_UNREGISTERED)) { (taskStatus.GetSource() == mesos.TaskStatus_SOURCE_SLAVE && taskStatus.GetReason() == mesos.TaskStatus_REASON_EXECUTOR_UNREGISTERED) ||
(taskStatus.GetSource() == mesos.TaskStatus_SOURCE_EXECUTOR && taskStatus.GetMessage() == messages.ContainersDisappeared)) {
//-- //--
// pod-task has metadata that refers to: // pod-task has metadata that refers to:
// (1) a task that Mesos no longer knows about, or else // (1) a task that Mesos no longer knows about, or else
// (2) a pod that the Kubelet will never report as "failed" // (2) a pod that the Kubelet will never report as "failed"
// (3) a pod that the kubeletExecutor reported as lost (likely due to docker daemon crash/restart)
// For now, destroy the pod and hope that there's a replication controller backing it up. // For now, destroy the pod and hope that there's a replication controller backing it up.
// TODO(jdef) for case #2 don't delete the pod, just update it's status to Failed // TODO(jdef) for case #2 don't delete the pod, just update it's status to Failed
pod := &task.Pod pod := &task.Pod