dynatrace

How to un/deploy Dynatrace

Default settings

By default your instance will be monitored with the Dynatrace credentials you provisioned (tenant, api_token, data_ingest_token) which are required when you monitor a Kubernetes Cluster with Dynatrace.

The default deployment mode is Classic full-stack injection but you can easily swap to the CloudNative full-stack injection.

Dynatrace credentials

The Dynatrace credentials will be passed along during the provisioning and are setted in the k8splay-install.sh file.

# ---- Define Dynatrace Environment ----
# Sample: https://{your-domain}/e/{your-environment-id} for managed or https://{your-environment-id}.live.dynatrace.com for SaaS
TENANT=
APITOKEN=
INGESTTOKEN=

You can find the above section at the top of the file.

Dynatrace Token Scopes

For ease of deployment is recommended you use the Dynatrace deployment UI Deployment > Start Installation > Kubernetes and you copy/paste:

APITOKEN="Generated Dynatrace Operator token" INGESTTOKEN="Generated Data ingest token"

💾 Storage of the credentials during/after provisioning

The Dynatrace credentials are saved in a configmap during installation in the default namespace. This allows a clean reinstallation even after deleting the namespace dynatrace so you can swap the monitoring mode.

    kubectl create configmap -n default dtcredentials --from-literal=tenant=${DT_TENANT} --from-literal=apiToken=${DT_API_TOKEN} --from-literal=dataIngestToken=${DT_INGEST_TOKEN}

Note: The variables are shuffled (e.g. TENANT -> DT_TENANT,...) inside the functions.sh file for reusability of functions and components.

🔄 Swap deployment mode and work with the deployment functions

If you want to swap the deployment mode from Classic full-stack to CloudNative full-Stack, just navigate to this directory

cd ~/k8s-play/cluster-setup/resources/dynatrace/

and load the deployment functions into your shell

source deploy_functions.sh

You'll see the message:

the functions deployOperator deployClassic deployCloudNative undeployDynakubes uninstallDynatrace have been loaded to the current shell

This means that as long as you don't change the shell, you can execute this functions. Then run

undeployDynakubes

to delete the dynakubes of the actual installation (will delete the OneAgent, ActiveGates and CSI Drivers).

Before you deploy the different deployment mode run kubectl get pod -n dynatrace to verify that the oneagent has been terminated succesfully (if not you can --force the deletion but this is normally not necessary).

After the undeployment is done you can then deploy the CloudNative full-stack deployment by running the command

deployCloudNative

Complete uninstall of Dynatrace

Run

source deploy_functions.sh

and then

uninstallDynatrace

Fresh install of Dynatrace with new Credentials

Run source credentials.sh and then

saveReadCredentials $DT_TENANT $DT_API_TOKEN $DT_INGEST_TOKEN

This will call the function saveReadCredentials and will override the credentials saved as configmap (if any) in the default namespace. You can run the saveReadCredentials function to verify the values of the configmap beforehand.

After you have setted your new credentials then run the function source deploy_functions.sh && deployOperator to deploy the operator and then deploy the deployment mode of your liking (either Classic or CloudNative).