AAP-46484 Capture k8s pod diagnostics #1332
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -29,4 +29,5 @@ venv | |
| /vendor | ||
| pkg/services/.lock | ||
| pkg/workceptor/status | ||
| pkg/workceptor/status.lock | ||
| **/__debug_* | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,139 @@ | ||
| package workceptor | ||
|
|
||
| import ( | ||
| "context" | ||
| "fmt" | ||
|
|
||
| corev1 "k8s.io/api/core/v1" | ||
| metav1 "k8s.io/apimachinery/pkg/apis/meta/v1" | ||
| "k8s.io/apimachinery/pkg/watch" | ||
| "k8s.io/client-go/kubernetes" | ||
| ) | ||
|
|
||
| type KubePodStateHelper interface { | ||
| CapturePodStatus(pod *corev1.Pod, stdoutSize int64, timeoutSeconds *int64) (ok bool, err error) | ||
| GetPodStatus(pod *corev1.Pod) (bool, string, error) | ||
| PodContainerHealthy(pod *corev1.Pod, containerName string) (bool, error) | ||
| PodHealthy(pod *corev1.Pod, containerName string) (bool, string, error) | ||
| WaitForPodCompleted(pod *corev1.Pod, clientset kubernetes.Interface, timeoutSeconds *int64) (*corev1.Pod, error) | ||
| } | ||
|
|
||
| func (kw *KubeUnit) CapturePodStatus(pod *corev1.Pod, stdoutSize int64, timeoutSeconds *int64) (bool, error) { | ||
| if pod == nil { | ||
| return false, fmt.Errorf("pod is nil") | ||
| } | ||
|
|
||
| if pod.Status.Phase == corev1.PodRunning || pod.Status.Phase == corev1.PodPending { | ||
| err := kw.WaitForPodCompleted(kw.GetContext(), pod, kw.clientset, timeoutSeconds) | ||
| if err != nil { | ||
| kw.GetWorkceptor().nc.GetLogger().Debug("Pod error detected while waiting for completion: %v", err) | ||
|
lranjbar marked this conversation as resolved.
Outdated
|
||
| } | ||
| } | ||
|
|
||
| ok, err := kw.GetPodStatus(pod) | ||
| if !ok || err != nil { | ||
| kw.GetWorkceptor().nc.GetLogger().Warning("Pod did not succeed: %v", err) | ||
| kw.UpdateBasicStatus(WorkStateFailed, err.Error(), stdoutSize) | ||
|
|
||
| return false, err | ||
| } | ||
|
|
||
| kw.GetWorkceptor().nc.GetLogger().Debug("Pod status: %s", pod.Status.String()) | ||
|
|
||
| return true, nil | ||
| } | ||
|
|
||
| // GetPodStatus checks if the pod has successfully completed its application logic and infrastructure is healthy. | ||
| func (ku KubeUnit) GetPodStatus(pod *corev1.Pod) (bool, error) { | ||
| if pod == nil { | ||
| return false, fmt.Errorf("pod is nil") | ||
| } | ||
|
|
||
| podRef := fmt.Sprintf("pod %s/%s", pod.Namespace, pod.Name) | ||
|
lranjbar marked this conversation as resolved.
Outdated
|
||
|
|
||
| ok, err := ku.PodHealthy(pod, containerName) | ||
| if !ok || err != nil { | ||
| return ok, fmt.Errorf("%s %s", podRef, err) | ||
| } | ||
|
|
||
| return ok, nil | ||
| } | ||
|
|
||
| // PodContainerHealthy checks if the pod has successfully completed its application logic. | ||
| // this is called after podInfrastructureSuccess has confirmed the pod is in a terminal state. | ||
| func (ku KubeUnit) PodContainerHealthy(pod *corev1.Pod, containerName string) (bool, error) { | ||
| if pod == nil { | ||
| return false, fmt.Errorf("pod is nil") | ||
| } | ||
|
|
||
| for _, cs := range pod.Status.ContainerStatuses { | ||
| if cs.Name == containerName { | ||
| if cs.State.Terminated == nil { // means it is waiting or running, so application logic has not completed yet. Normal behavior when job completes successfully. | ||
| return true, nil | ||
| } | ||
|
|
||
| if cs.State.Terminated.ExitCode != 0 { // exit code of 0 means success | ||
| return false, fmt.Errorf("container %s exited with code %d: %s", cs.Name, cs.State.Terminated.ExitCode, cs.State.Terminated.Reason) | ||
| } | ||
|
|
||
| return true, nil // container terminated with exit code of 0 | ||
| } | ||
| } | ||
|
|
||
| return false, fmt.Errorf("pod %s/%s does not contain container %s", pod.Namespace, pod.Name, containerName) | ||
|
lranjbar marked this conversation as resolved.
Outdated
|
||
| } | ||
|
|
||
| // PodInfrastructureSuccess checks if the pod has either successfully started, is pending or is running, or has is successfully terminated. | ||
| // Any other state is considered an infrastructure failure. | ||
| func (ku KubeUnit) PodHealthy(pod *corev1.Pod, containerName string) (bool, error) { | ||
| if pod == nil { | ||
| return false, fmt.Errorf("pod is nil") | ||
| } | ||
|
|
||
| for _, cs := range pod.Status.ContainerStatuses { | ||
| // Check if this is the container we care about first | ||
| if cs.Name == containerName { | ||
| switch pod.Status.Phase { | ||
| case corev1.PodFailed: | ||
| ok, err := ku.PodContainerHealthy(pod, containerName) | ||
| if !ok || err != nil { | ||
|
|
||
| return false, err | ||
| } | ||
|
|
||
| return true, nil | ||
| case corev1.PodSucceeded: | ||
| return true, nil | ||
| case corev1.PodRunning, corev1.PodPending: | ||
| return true, nil | ||
| default: | ||
| return false, fmt.Errorf("unknown phase: %s", pod.Status.Phase) | ||
| } | ||
| } | ||
| } | ||
|
|
||
| return false, fmt.Errorf("pod %s/%s does not contain container %s", pod.Namespace, pod.Name, containerName) | ||
|
lranjbar marked this conversation as resolved.
Outdated
|
||
| } | ||
|
|
||
| func (ku KubeUnit) WaitForPodCompleted(ctx context.Context, pod *corev1.Pod, clientset kubernetes.Interface, timeoutSeconds *int64) error { | ||
| if pod == nil { | ||
| return fmt.Errorf("pod is nil") | ||
| } | ||
|
|
||
| watcher, err := clientset.CoreV1().Pods(pod.Namespace).Watch(ctx, metav1.ListOptions{ | ||
| TimeoutSeconds: timeoutSeconds, | ||
| FieldSelector: "involvedObject.kind=Pod,involvedObject.name=" + pod.Name}) | ||
| if err != nil { | ||
| return err | ||
| } | ||
|
|
||
| for event := range watcher.ResultChan() { | ||
| if event.Type == watch.Error { | ||
| return event.Object.(error) | ||
| } | ||
| pod = event.Object.(*corev1.Pod) | ||
| fmt.Printf("%s: %s/%s (Phase: %s)\n", event.Type, pod.Namespace, pod.Name, pod.Status.Phase) | ||
|
lranjbar marked this conversation as resolved.
Outdated
|
||
| } | ||
|
|
||
| return nil | ||
| } | ||
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.