dynatrace-bootstrapper

The dynatrace-bootstrapper is a small CLI binary built into a Dynatrace CodeModule (after version <to-be-determined-codemodule-version>) so that it can be used in conjunction with the dynatrace-operator to optimize the download and configuration of a CodeModule in Kubernetes.

Responsibilities

• Copy a Dynatrace CodeModule to a target directory
• Configure the Dynatrace CodeModule according to the configuration options provided

How to use

Args

--source

Example: --source="/opt/dynatrace/oneagent"

• ⚠️This is a required arg⚠️
• The --source arg defines the base path where to copy the CodeModule FROM.

--target

Example: --target="example/bins/1.2.3"

• ⚠️This is a required arg⚠️
• The --target arg defines the base path where to copy the CodeModule TO.

--work

Example: --work="/example/work"

• This is an optional arg
• The --work arg defines the base path for a tmp folder, this is where the command will do its work, to make sure the operations are atomic. It must be on the same disk as the target folder.

--technology

Example: --technology="python,java"

• This is an optional arg
• The --technology arg defines the paths associated to the given technology in the <source>/manifest.json file. Only those files will be copied that match the technology. It is a comma-separated list.

--config-directory

Example: --config-directory="/example/config/dir"

• This is an optional arg
• The --config-directory arg defines the path where enrichment and config files will be created in.

--input-directory

Example: --input-directory="/example/input"

• This is an optional arg
• The --input-directory arg defines the base path where configuration files are provided.
  • Intended use is to mount a k8s Secret with the correct content in the provided path.
    • So when we are talking about "files", it is just the fields of the Secret's data
  • Config files:

    • ruxitagentproc.json: A json file containing a response from the /deployment/installer/agent/processmoduleconfig endpoint of the Dynatrace Environment(v1) API.

      • This file is required if --input-directory is defined.
      • Used to create the <config-directory>/<container-name>/oneagent/agent/config/ruxitagentproc.conf file

    • initial-connect-retry: A file containing a single number value. Defines the delay before the initial connection attempt. (Useful in case of istio-proxy is used.)

      • Used to create/update the <config-directory>/<container-name>/oneagent/agent/customkeys/curl_options.conf file.

    • trusted.pem: A file containing the certificates used by the CodeModule for all its communication (proxy communication's not included).

      • Used to create the <config-directory>/<container-name>/oneagent/agent/customkeys/custom.pem and <config-directory>/<container-name>/oneagent/agent/customkeys/custom_proxy.pemfile.

    • activegate.pem: A file containing the certificates used by the CodeModule for all its communication with the ActiveGate (proxy communication's NOT included).

      • Used to create the <config-directory>/<container-name>/oneagent/agent/customkeys/custom.pem.
      • Is concatenated with the trusted.pem if both is present.

    • endpoint.properties: A file containing the necessary info so the metadata-enrichment metrics can be ingested properly

      • Used to create the <config-directory>/<container-name>/enrichment/endpoint/endpoint.properties.
      • Example:

      DT_METRICS_INGEST_URL=http://test-activegate.dynatrace/e/tenant/api/v2/metrics/ingest
      DT_METRICS_INGEST_API_TOKEN=<data-ingest-token>
      

--install-path

Example: --input-directory="/opt/dynatrace/oneagent"

• This is an optional arg
  • Defaults to /opt/dynatrace/oneagent
• The --install-path arg defines the base path where the agent binary will be put. This is only necessary to properly configure the ld.so.preload file.
  • The ld.so.preload is put under <config-directory>/oneagent/ld.so.preload

--fullstack

Example: --fullstack

• This is an optional arg
  • Defaults to false
• The --fullstack arg will make sure that the CodeModule is configured to be in fullstack mode.
  • Adds additional values to the <config-directory>/<container-name>/oneagent/agent/config/container.conf.

--tenant

Example: --tenant="my-tenant"

• This is an optional arg, but mandatory incase of --fullstack.
• Only used incase of --fullstack, provides additional info needed to properly configure <config-directory>/<container-name>/oneagent/agent/config/container.conf.

--attribute

Example: --attribute="k8s.pod.name=test"

• This is an optional arg
• The --attribute arg defines the passed in Pod attributes that will be used to configure the metadata-enrichment and injected CodeModule. It is a key value pair.

--attribute-container

Example: --attribute-container="{k8s.container.name=app, container_image.registry=gcr.io, container_image.repository=test}"

• This is an optional arg
• The --attribute-container arg defines the passed in Container attributes that will be used to configure the metadata-enrichment and injected CodeModule. It is a JSON formatted string.

--suppress-error

Example: --suppress-error

• This is an optional arg
  • Defaults to false
• The --suppress-error arg will silence any errors, causing the executable to return with an exit code 0 even if an error occurred. Intended purpose is to not block the application container from starting, if used as init-container.

--debug

Example: --debug

• This is an optional arg
  • Defaults to false
• The --debug arg will enabled the debug logs.

Development

• To run tests: make test
• To run linting: make lint
• To build a testing image: make build

Debug logs

• The flag --debug enables debug logs.
• Create new debug logs via log.V(1).Info(...)
  • The logging levels between zap and logr is different, so we had to get creative.

How to test (in K8s)

helm-sample

A a Helm chart for a small PHP sample app, with the bootsrapper as it's initContainer, as it is expected to be used.

To deploy it in your cluster (where your KUBECONFIG is pointing) with the image of the current branch (considering that the image was built via make build):

• make deploy

To remove it from your cluster:

• make undeploy

Custom values

You can use the make deploy/custom

• This will use the hack/testing/helm-sample/_values.yaml as the values.
  • This file is ignored by git, so you can safely put whatever you want into it.
• The image will still be set according to what make build would create.

How to release

The bootstrapper is mainly used by installing it via go install, so the release is very simple:

1. Go to the release page.
2. Press Draft new release
3. Set the tag you want to release, you don't need to push the tag beforehand. Example format: v1.2.3
4. Press Generate release notes
5. Cleanup generated output to only have the interesting commits. (business logic relevant or bugfixes)
6. Publish release