Introduction

The Docker Integration Tests (aka BDI) is image for integration tests. Supposed that this image will not be used to execute integration tests directly but real images for integration tests will use this image as basic (FROM command in the particular docker file). BDI builds some sandbox which includes python interpreter, Robot Framework, some useful tools such as bash, shadow, vim, rsync, ttyd, common custom Robot Framework libraries (for example, PlatformLibrary) and customized docker entry point script.

Pre-installed tools

The Docker Integration Tests contains the following pre-installed Linux tools:

• python (version 3.10.13)
• bash
• shadow
• vim
• rsync
• ttyd

Library documentation

PlatformLibrary documentation is autogenerated by Robot Framework libdoc tool. It can be found here To generate new doc you should navigate to current project repository and execute the following command:

python -m robot.libdoc PlatformLibrary.py PlatformLibrary.html

and move PlatformLibrary.html file to documentation directory.

Docker Entry point Script

A docker entry point script is a script which will be executed after docker container is created. If you override the image, its entry point script will be executed by default. But if you override the entry point as well, your own entry point will be run. Docker Integration Tests contains simple and customized entry point script - /docker-entrypoint.sh with the following command (possible CMD arguments):

• run-robot is a default CMD command which executes Robot Framework test cases, can resolve Robot Framework tags to be excluded, can create result.txt file with parsed Robot Framework tests results in pretty format. After tests execution ttyd tool is started.

• run-ttyd command starts ttyd tool. ttyd is Web-console which rather useful for dev and troubleshooting purposes.

• custom command executes custom bash script if this script's path is provided. To provide custom script this script should exist within container and environment variable CUSTOM_ENTRYPOINT_SCRIPT should contain path to the script. Actually, custom command is equivalent to overriding the entry point script but we recommend implementing custom script instead of entry point overriding.

Example of equivalent console command:

/docker-entrypoint.sh run-robot

Below is detailed description regarding run-robot command.

run-robot description

run-robot command contains 4 customized steps:

• service checker script. Sometimes we want to make sure that "tested" service is ready for testing and start execution only for "ready" service. For this purpose inheritor image should implement "service checker script" (python) which gives timeout argument in seconds (by default timeout is 300 second but can be overridden by SERVICE_CHECKER_SCRIPT_TIMEOUT environment variable) and checks if the service is ready. If the service is ready entry point script goes to the next step. To specify "service checker script" SERVICE_CHECKER_SCRIPT environment variable should not be empty and should contain path to custom python script. By default, SERVICE_CHECKER_SCRIPT is empty and current entry point script step is skipped.

• excluded tags resolver. Sometimes some Robot Framework tests can not be executed in given configuration. For example, test should check Elasticsearch service but elasticsearch URL is not given and we want to skip this test. The Robot Framework paradigm supposed that this case should be resolved by tag approach. We can point which tests should be skipped by its tags. For example, robot -e my-excluded_tag ./tests. Docker Integration Tests entry point script provides development approach to recognize which tests should be skipped before tests execution. Supposed that .robot files are contained some root folder (for, example tests). robot_tags_resolver.py script recursively bypasses all inner folder to look up all tags_exclusion.py modules. For each found module get_excluded_tags(environ) function will be executed, where where environ is a list of environment variables. get_excluded_tags function should return list or dictionary of excluded tags. If list is returned, all tags will be added to result set of excluded tags. If dictionary is returned, all keys will be added to result set of excluded tags and the particular dictionary will update result dictionary which will be printed to console as some map where keys are excluded tags and values are reasons why these tags are excluded. Default tags resolver script is robot_tags_resolver.py but inheritor image can override it by TAGS_RESOLVER_SCRIPT environment variable which contains path to custom tags resolver script. To skip excluded tags resolving process environment variable IS_TAGS_RESOLVER_ENABLED should be false (it is true by default).

• robot tests execution. This step can not be skipped. It executes Robot Framework tests without excluded ones. If TAGS environment variable is presented only these tags will be executed. This is an example of specifying only first and second tests from first, second and third:

firstORsecond

If tag of test is contained in included and excluded tags it will not be executed.

• analyze results. Sometimes text file with Robot Framework results should be generated in pretty human readable format. For example, we want to copy this result from Kubernetes Pod to Jenkins job and we want text results instead of all Pod's logs or html formatted file. For this purpose analyze_result.py module will be executed. This module creates result.txt file in output folder (with all Robot Framework results) with tests results in pretty format. To skip this step environment variable IS_ANALYZER_RESULT_ENABLED should be false (it is true by default). Default analyzer script is analyze_result.py but inheritor image can override it by ANALYZE_RESULT_SCRIPT environment variable which contains path to custom analyzer script.

• write status. To integrate with deployer Jenkins Job status of integration tests should be set to watched by Jenkins Job Kubernetes entity. It can be as custom resource (CR) as native Kubernetes entities (deployments, pods, etc.). The BDI provides an ability to write status of executed tests to some Kubernetes entity out of the box. Status is written as Kubernetes status condition with the following fields:

• lastTransitionTime - timestamp.
• message - parsed results of Robot framework integration tests as string.
• reason - static field with IntegrationTestsExecutionStatus value.
• status - can be True or False. This field depends on type field. It is True if type is Ready or Successful and False if type is Failed or In progress.
• type - can be Ready, Successful, Failed or In progress.

For example,

lastTransitionTime: "2021-04-21T11:21:31.332Z"
message: <Robot Framework parsed results>
reason: IntegrationTestsExecutionStatus
status: "True"
type: Ready

In some cases, it is necessary to have status field as boolean instead of string.

For example,

lastTransitionTime: "2021-04-21T11:21:31.332Z"
message: <Robot Framework parsed results>
reason: IntegrationTestsExecutionStatus
status: true
type: Ready

To support this, environment variable IS_STATUS_BOOLEAN must be set to true. By default, IS_STATUS_BOOLEAN is considered to be false.

Note! For using feature write status in restricted environment, the user or service account used by the Deployer should have permissions on entity group with the verbs get, patch and resources <resource_plural>/status in current the namespace or project. For example, permissions for write status in Custom Resource:

- apiGroups:
  - qubership.org
  resources:
  - platformmonitorings/status
  verbs:
  - get
  - patch

To write status to some k8s entity you should specify the entity. There are two ways to do this. The first one is to specify full path. For example, you have ZooKeeperService custom resource which has metadata.selfLink field with value - /apis/qubership.org/v1/namespaces/zookeeper-service/zookeeperservices/zookeeper, in the current approach you should specify STATUS_CUSTOM_RESOURCE_PATH environment variable with value from selfLink without apis prefix and namespaces part:

STATUS_CUSTOM_RESOURCE_PATH=qubership.org/v1/zookeeper-service/zookeeperservices/zookeeper

The second approach is to point the path in parts using the following environment variables:

STATUS_CUSTOM_RESOURCE_GROUP=qubership.org
STATUS_CUSTOM_RESOURCE_VERSION=v1 
STATUS_CUSTOM_RESOURCE_NAMESPACE=zookeeper-service
STATUS_CUSTOM_RESOURCE_PLURAL=zookeeperservices 
STATUS_CUSTOM_RESOURCE_NAME=zookeeper

If your k8s pod with integration tests always writes status to well-known custom resource you can override all this environment variables (excluding STATUS_CUSTOM_RESOURCE_NAMESPACE) in your docker file and set namespace in helm chart.

Both of this approaches work with native k8s entities too. For example,

STATUS_CUSTOM_RESOURCE_GROUP=apps
STATUS_CUSTOM_RESOURCE_VERSION=v1
STATUS_CUSTOM_RESOURCE_NAMESPACE=zookeeper-service
STATUS_CUSTOM_RESOURCE_PLURAL=deployments
STATUS_CUSTOM_RESOURCE_NAME=zookeeper-1

If feature is available write_status.py script is called two times. The first time immediately after docker entrypoint script was started to set In progress condition. The second time after tests are finished and parsed by analyze results script to set in the message field tests results. Default analyzer script is write_status.py but inheritor image can override it by WRITE_STATUS_SCRIPT environment variable which contains path to custom "write status" script.

By default, if all tests are passed BDI set Ready value to type condition field. There is an ability to deploy only integration tests without any component (component was installed before). In this case you should set ONLY_INTEGRATION_TESTS environment variable as true and BDI will set Successful as value of type condition field.

The message field in the status condition by default contains first line from result.txt file (which is generated in the previous step). To write full parsed result you should set IS_SHORT_STATUS_MESSAGE environment variable to false.

Important! If you use custom script to parse result (ANALYZE_RESULT_SCRIPT is not empty) please pay attention that result should be placed in the result.txt file and the first line will be used as short status message.

Note! This feature (write status to k8s entities) is disabled by default! To turn on it please set the STATUS_WRITING_ENABLED environment variable to true. For example in your docker file as

ENV STATUS_WRITING_ENABLED=true

Environment Variables

Docker Integration Tests uses the following environment variables:

• DEBUG
• TTYD_PORT
• CUSTOM_ENTRYPOINT_SCRIPT
• SERVICE_CHECKER_SCRIPT
• SERVICE_CHECKER_SCRIPT_TIMEOUT
• IS_TAGS_RESOLVER_ENABLED
• IS_ANALYZER_RESULT_ENABLED
• ANALYZE_RESULT_SCRIPT
• STATUS_CUSTOM_RESOURCE_GROUP
• STATUS_CUSTOM_RESOURCE_VERSION
• STATUS_CUSTOM_RESOURCE_NAMESPACE
• STATUS_CUSTOM_RESOURCE_PLURAL
• STATUS_CUSTOM_RESOURCE_NAME
• ONLY_INTEGRATION_TESTS
• STATUS_CUSTOM_RESOURCE_PATH
• STATUS_WRITING_ENABLED
• WRITE_STATUS_SCRIPT
• IS_SHORT_STATUS_MESSAGE
• TAGS
• IS_STATUS_BOOLEAN

All of them instead of TAGS, ONLY_INTEGRATION_TESTS, STATUS_CUSTOM_RESOURCE_NAMESPACE, STATUS_CUSTOM_RESOURCE_PATH and maybe DEBUG we recommend overriding in the docker file and do not forward them to the integration tests deployment environment.