WRONG: 3:01 Minutes audio => "...if the application is not responsive, liveness probe will be kicked in and only the application i.e. container will be restarted, the Pod won't get recreated. But in terms of readiness, if readiness fails, the Pod will be recreated" CORRECT: (1) If readinessProbe fails, the EP controller removes Pod'S IP address from EP of all services that match the Pod. (Does not restart Pod as this video explains). So readinessProbe is a mechanism to get traffic only when the application is ready. If not ready, do not serve traffic. (2) If livenessProbe fails, the kubelet kills the Container, and the Container is subjected to Pod restartPolicy (applied to all containers of Pod) is set to Always or OnFailure. Refer: kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/ kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/ kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/#restart-policy
You can leverage "failureThreshold" config applied on your probes. Here is a sample from k8s docs - kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-startup-probes
Can we run pod without Liveliness & readiness probes. As application startup is taking 5 mins and due to Liveliness pod is getting restarted all the times
assume ,i hav 3 enquiry rest service and one pub-sub service for transaction, and one listener which execute batch.what would be my deployment strategy for production?pls advice...
![[ Kube 79 ] Kubernetes Liveness & Readiness Probes](http://i.ytimg.com/vi/3TJRkKWuVoM/mqdefault.jpg)
Thanks man , that's the best explanation 👍
WRONG: 3:01 Minutes audio => "...if the application is not responsive, liveness probe will be kicked in and only the application i.e. container will be restarted, the Pod won't get recreated. But in terms of readiness, if readiness fails, the Pod will be recreated"
CORRECT: (1) If readinessProbe fails, the EP controller removes Pod'S IP address from EP of all services that match the Pod. (Does not restart Pod as this video explains). So readinessProbe is a mechanism to get traffic only when the application is ready. If not ready, do not serve traffic.
(2) If livenessProbe fails, the kubelet kills the Container, and the Container is subjected to Pod restartPolicy (applied to all containers of Pod) is set to Always or OnFailure.
Refer:
kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/
kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/
kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/#restart-policy
verry clear Thanks,
Could you please upload your Bangalore session on kubernates hands on? Thanks for everything you do for us.
The content is exactly the same which you see here. The meetups were in-person events and I just was speaking in one of the topic :)
Why is sf5hf container not terminating after 2 minutes? Shouldn't liveness probe kick in and terminate it?
Amit Padgaonkar it waits until the failureThreshold is reached. By default its 3
Hi Ajay..thnx for the video . 1 qq any idea how to set the max threshold for pod restart incase liveness keep failing?
You can leverage "failureThreshold" config applied on your probes. Here is a sample from k8s docs - kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-startup-probes
Can we run pod without Liveliness & readiness probes. As application startup is taking 5 mins and due to Liveliness pod is getting restarted all the times
assume ,i hav 3 enquiry rest service and one pub-sub service for transaction, and one listener which execute batch.what would be my deployment strategy for production?pls advice...
are these all separate microservices? or single monolithic service?
@@TechPrimers only 3 enquiry services are ms. rest monolithic,listener should be a single instance.