Hello, We are in a process of using this new Dynatrace Cloud Native deployment to monitor my Kubernetes Cluster.
Right now we use Classic Full Stack and there is a limitation of one agent Pods to be running first so that we can get the deep level monitoring of application pods otherwise everytime we would have to restart the app Pods.
Apparently the Cloud Native deployment has come up with a solution to overcome this limitation. BUT it seems like it is also bringing a very hard dependency on One agent Pods i.e. during the node scaling app is waiting on the One agent and csi-driver pods to come up first which concern me with 2 things -
- It is adding a additional start up time for app pods because they are waiting for one agent pods to come up which is alright but the main concern is the 2nd point below.
- Since the app is waiting for Dynatrace Pods to come up, what happens when for some reason Dynatrace pods doesn't come up? Does that mean the app will keep waiting which is obviously not ideal in Production. Is there a way to set a timeout or something where in case Dynatrace Pods are not coming up, app pods will wait for some time and then ignore it? This is really important because it's not ideal to create a hard dependency on Dynatrace Pods and creating a choke point.
Let me know if anyone has any advice on this please.
Best Regards,
Shashank
