Exam Professional Cloud Architect topic 1 question 122 discussion

According to the reference, answer should be A. https://cloud.google.com/blog/products/management-tools/using-logging-your-apps-running-kubernetes-engine

IMHO a) is the correct answer, not c) The point is, that we have a scenario in that often errors in GKE happen - within 2 week a lot of people complained about a lot of errors. For the past we have no data at all as we have not monitored anything. That means we will collect data from now on to find out what the problem is. The additional value of an alert is not clear - and it for me not clear why we need additionally to install Prometheus considering that until now we had no GKE monitoring at all. Please correct me if I am wrong.

B & D are asking to create new GKE cluster so we can ignore them From A and C , we can achieve the requirement with the help of A itself. Monitoring & Logging should help investigate the issue.

C 1. "You currently have no logging or monitoring solution enabled on your GKE cluster" and "you have not been able to replicate the issue"- nothing interesting in GKE monitoring dashboard 2. No alerting in answer A

Options A and C are less disruptive. Option C adds Prometheus on top which looks like overkills for this simple/initial level of troubleshooting. I would go with option A

A A provides native solution to GCLoud Why not C? from GCP best practices for GKE we should rely on native logging capabilities. No need for additional solutions like Prometheus. Also it is about reviewing logs, monitoring the service, not receiving alerts each time its happens, that will not provide any insight on the issue. Prometheus could potentially help identify when the issue occurs, it doesn't directly help with diagnosing the root cause of the problem. B & D rejected because migration will cause distruption. https://cloud.google.com/blog/products/management-tools/using-logging-your-apps-running-kubernetes-engine

If we enable Cloud Operations we should be able to see the logs from this point onwards. Data of past errors would not be visible. It's not rational to expect developers to check every hour for appearances of the error in the logs, and that's where an alert comes in handy. It'll notify you when the conditions that led to the error appear again so that developers can analyze the logs and understand the problem. I agree that installing Prometheus is not needed today, but it seems that it was the only option at the time they created this question to set up alerts and, in my opinion, the alerts are vital to diagnose the problem.

Marked in Green is the real exam ans, or the community most voted one? I'm confused now hehe

As per https://cloud.google.com/stackdriver/docs/managed-prometheus - Correct option, I feel is C.

Answer is A

I think answer A is enough, but if you want a more complete solution C could be a good option: https://cloud.google.com/stackdriver/docs/managed-prometheus

Could anyone kindly explain why B is incorrect? Thank you.

A. 1. Update your GKE cluster to use Cloud Operations for GKE. 2. Use the GKE Monitoring dashboard to investigate logs from affected Pods. By updating your GKE cluster to use Cloud Operations for GKE (formerly known as Stackdriver), you enable monitoring and logging without disrupting the application. The GKE Monitoring dashboard allows you to investigate logs from affected Pods, which helps you diagnose the problem that customers have reported. This approach minimizes disruption to the application while providing the necessary information to identify and resolve the issue

As described here https://cloud.google.com/stackdriver/docs/solutions/gke is it possible to install prometheus as part of cloud operation suite.

you want to minimize change so A is the best answer also you don't know what is the issue by reviewing the logs you can find something

The best option is D. 1. Create a new GKE cluster with Cloud Operations for GKE enabled, and deploy Prometheus. 2. Migrate the affected Pods to the new cluster, and redirect traffic for those Pods to the new cluster. 3. Set an alert to trigger whenever the application returns an error. Here's why: Option A does not involve creating a new GKE cluster, which means you will not be able to isolate the affected Pods from the rest of the application. This can make it difficult to diagnose the issue without disrupting the entire application.

Question states problem cannot be replicated. So alerting is required to review the right logs at right time. Hence A is not adequate solution and C is the right one