Migrate from GCP function 1.0 to GCP k8s 1.0
Dynatrace version 1.230+
Since function-based GCP integration (v1.0) is being deprecated, users are expected to migrate to k8s-based GCP integration (v1.0).
Both kinds of integration have the same features. Only the deployed components change. Data continuity is preserved, except for the time elapsed between deployments. You don't need to change anything in your dashboard or alerts as metric keys stay the same.
Before you begin
It is recommended that you have the activation-config.yaml file that you used during the function deployment. Most of the parameters you need for a new deployment can be reused based on this file.
The activation-config.yaml file specifies the monitoring scope (services and feature sets). You need to specify the same monitoring scope in your new deployment to confirm that monitoring coverage does not change.
If you don't have the file, you can check the current monitoring scope in function deployment:
-
Find your monitoring function in GCP. By default, its name is
dynatrace-gcp-monitor. -
Go to Variables tab.
-
Copy
ACTIVATION_CONFIGvariable and save it. You can use it as reference when setting monitoring scope (services and feature sets) for new deployment.
Migration guide
-
Remove function-based GCP integration.
Rolling upgrade (deploying a new integration before removing the old one) is not recommended, as the metrics would be reported twice, which might affect your data.
See Uninstall.
Make sure that you have disabled all GCP extensions in Dynatrace UI. Alternatively, you can add
--upgrade-extensionsoption during deployment in step 3 when running./deploy-helm.shdeployment script. -
Deploy k8s-based GCP integration from scratch.
See Set up the Dynatrace GCP metric and log integration on a new GKE Autopilot cluster.
-
Verify that k8s-based GCP integration is working.
See Verify installation.