Skip to content

Latest commit

 

History

History
110 lines (86 loc) · 3.9 KB

developers-notes.md

File metadata and controls

110 lines (86 loc) · 3.9 KB
Raw

Developer's Guide & Tips

Running a custom operator

As noted in the README the operator can be deployed using a custom image. This section elaborates on that.

The makefile is aware of two variables (env vars or directly used by make):

  • TAG - specify a custom tag for your container image
  • IMG - specify a custom image (repository & tag) for your image

In the following examples, we assume you will be testing using your own container repository and thus will use a fully specified IMG variable.

# set the IMG var for subsequent commands
export IMG=registry.example.com/myuser/samba-operator:test
# build the container image
make image-build
# push the image to a container registry
make container-push
# populate k8s with CRDs and launches the operator.
# assumes kubectl is set up and works
make deploy

Behind the scenes this makefile uses kustomize and loading resources into the kubernetes cluster is handled via the YAML files in ./config. There is a special makefile target set-image that runs kustomize commands in order to set a YAML file in that directory to use your container image rather than a default one. This target is automatically used by make deploy but can be used manually if needed.

Please do not check changes made by kustomize to kustomization.yaml files in to git history.

Testing with a custom operator

To verify the test scripts are testing the right image, a rule checks that the deployed operator in the kubernetes cluster matches what the test expects. The test's expectation is controlled via an environment variable SMBOP_TEST_EXPECT_MANAGER_IMG. To ensure the tests match the custom container image you used this variable should also be set. Example:

# configure the tests to check for a given container
export SMBOP_TEST_EXPECT_MANAGER_IMG="${IMG}"
# Run the tests
./tests/test.sh

Using a custom samba server container image

The operator itself will create pods running various samba-server container images. Certain aspects of the operator, such as the container image to use for the samba server are configurable. There are a few ways to configure the operator, as it can read it's config from TOML or YAML files as well as it's command line or environment variables. The following example uses environment variables set in the operator's own pod spec.

We will set the environment variables using kustomize rules in the file ./config/manager/kustomization.yaml. Add the following to that file:

patches:
- patch: |-
    - op: add
      path: /spec/template/spec/containers/0/env/-
      value:
        name: "SAMBA_OP_SMBD_CONTAINER_IMAGE"
        value: "registry.example.com/myuser/samba-server:experiment"
  target:
    kind: Deployment

For multiple values, more than the "op" in the embedded yaml patch can be specified. Using "add" with "/spec/template/spec/containers/0/env/-" means to append the value to the end of the list at "env".

You can also add your own "kustomize' patches and other rules. See the kustomize docs for more information on using kustomize. You can also set other environment variables in a similar manner.

Please do not check changes made by kustomize to kustomization.yaml files in to git history.

Debugging the samba containers

Similar to using a custom container image the operator accepts a configuration value for samba debugging that will be passed on to the containers the operator creates. This parameter is samba-debug-level in configuration files and SAMBA_OP_SAMBA_DEBUG_LEVEL in the evnironment. The value should be a numeral 0 through 10 specified as a string.

Example setting the variable via the ./config/manager/kustomization.yaml file:

patches:
- patch: |-
    - op: add
      path: /spec/template/spec/containers/0/env/-
      value:
        name: "SAMBA_OP_SAMBA_DEBUG_LEVEL"
        value: "8"
  target:
    kind: Deployment