RPL-3.0-runner

local PROD integration testing via docker compose

IMPORTANT: Make sure to follow these setup instructions first: RPL-3.0 (Backend APIs): local PROD via docker compose

ONLY AFTER the metaservices and the APIs from RPL-3.0 are running, you can run this to start both the receiver and the runner in a containerized environment:

docker compose -f docker-compose.local.yml up -d --build

To stop them:

docker compose -f docker-compose.local.yml down

You can spin up the frontend whenever you want, the only mandatory services are the backend ones (so that this compose file can find their networks and connect to them).

local PROD integration testing via minikube

IMPORTANT: Make sure to follow these setup instructions first: RPL-3.0 (Backend APIs): local PROD via minikube

ONLY AFTER the backend's minikube setup, you can follow these steps to start the kubernetes deployment and service:

• Build the docker images:

docker build -t worker_rpl:local . # (this is the receiver)
docker build -t runner_rpl:local ./rpl_runner

• Temporarily edit the kubernetes/deployments/runner.yaml file to use the worker_rpl:local image for the receiver and the runner_rpl:local image for the runner. Also you should comment out the following contents:

  nodeSelector:
    cloud.google.com/gke-preemptible: "true"
  tolerations:
  - effect: NoSchedule
    key: task
    operator: Equal
    value: preemptive

• Start the kubernetes deployment and service:

kubectl create -f kubernetes/deployments/runner.yaml
kubectl create -f kubernetes/services/runner.yaml

To stop the deployment or the service, you can run:

kubectl delete -f ./kubernetes/deployments/runner.yaml
kubectl delete -f ./kubernetes/services/runner.yaml

If you want to stop everything at once, you can run this: (WARNING: this will stop ALL deployments/services within the namespace, not just the runner related ones)

kubectl delete --all deployments --namespace=default
kubectl delete --all services --namespace=default

---

---

---

DEPRECATED: README.md content from RPL-2.0-runner:

This was mainly local testing of the single image for the runner and the receiver. Currently it is strongly recommended to use the previously mentioned kubernetes/docker-compose setup for local PROD integration testing.

USAGE

Build receiver docker image:

docker build -t rpl-2.0-runner ./rpl_runner

With rabbitmq

1.- Start a rabbitmq server

docker run -p 9999:15672 -p 5672:5672 deadtrickster/rabbitmq_prometheus:latest

2.- Start receiving messages with

python3 rabbitmq_receive.py

3.- Send messages with (or just POST a submission to the RPL-server)

python3 rabbitmq_send.py <submission_id> <lang>

WHERE LANG can be: python_3.7 or c_std11

Without rabbitmq

docker build -t rpl-2.0-runner ./rpl_runner && python3 receiver.py <submission_id> <lang>

Description

rabbitmq_send.py

ONLY FOR TESTING: Send a message to the queue

rabbitmq_receive.py

Receive message from queue and invoque receiver.py

receiver.py

Get submission from RPL-Server Creates NEW DOCKER PROCESS with a custom runner for a specific language in a docker container Post results to RPL

init.py, runner.py and custom_runner.py (RUNNING INSIDE DOCKER CONTAINER)

1. Prepare
2. Build
3. Run

Every step is language-specific and redirecting stdout and stderr to a tempfile.

stdout result example:

{
  "tests_execution_result_status":"OK",                                                                                                                                                                                    
  "tests_execution_stage":"COMPLETE",
  "tests_execution_exit_message": "Completed all stages",
  "unit_test_suite_result_summary": {
    "amount_passed": 2,
    "amount_failed": 0,
  "amount_errored": 0,
  "single_test_reports":[
    {
      "name": "test_1",
      "status": "PASSED",
      "messages": null
    }
    {
      "name": "test_2",
      "status": "PASSED",
      "messages": null
    }                                                                                                                                         
  ],
  "tests_execution_stdout": "2020-07-19 20:16:55,844 RPL-2.0      INFO     Build Started\n2020-07-19 20:16:55,845 RPL-2.0      INFO     Building\n2020-07-19 20:16:55,846 RPL-2.0      INFO     start_BUILD\n/usr/bin/python3.7 -m py_compile  unit_test.py unit_test_wrapper.py assignment_main.py\n2020-07-19 20:17:02,219 RPL-2.0      INFO     end_BUILD\n2020-07-19 20:17:02,219 RPL-2.0      INFO     Build Ended\n2020-07-19 20:17:02,219 RPL-2.0      INFO     Run Started\n2020-07-19 20:17:02,221 RPL-2.0      INFO     Running Unit Tests\n2020-07-19 20:17:02,221 RPL-2.0      INFO     start_RUN\n\n2020-07-19 20:17:02,287 RPL-2.0      INFO     end_RUN\n2020-07-19 20:17:02,288 RPL-2.0      INFO     RUN OK\n2020-07-19 20:17:02,288 RPL-2.0      INFO     Run Ended\n",
  "tests_execution_stderr":"",
} 

How I tested it

In any case, you have to build the runner docker image that the receiver uses for testing assignments in a containerized environment.

docker build -t rpl-2.0-runner ./rpl_runner

Any changes made to any file inside the rpl_runner folder will need to rebuild the docker image.

The whole circle

Start the rabbitmq server, the Java RPL Server and the rabbitmq_receive.py

Login, get your auth token.

Post a submission like (replace /home/alepox/Desktop/tp_prof/rpl/pruebita_runner/la_submission.tar.xz and the Authorization header)

curl -X POST \
  {producer_base_api}/api/courses/1/activities/1/submissions \
  -H 'Accept: */*' \
  -H 'Accept-Encoding: gzip, deflate' \
  -H 'Authorization: Bearer eyJhbGciOiJIUzUxMiJ9.eyJzdWIiOiIxIiwiaWF0IjoxNTc4MTM5MjA0LCJleHAiOjE1NzgxNTM2MDR9.nSub-DaZq5bejmN_h31gymg4xay--mbSD0_ogpGOsDJzV1pP_gkSwCF_LnwfSm0UsLFYuuoCHruC2V8hCaaIqA' \
  -H 'Cache-Control: no-cache' \
  -H 'Connection: keep-alive' \
  -H 'Content-Length: 1025' \
  -H 'Content-Type: multipart/form-data; boundary=--------------------------193299411849957077408518' \
  -H 'Host: localhost:8080' \
  -H 'Postman-Token: 66b86c8a-1d39-44fc-b2e5-309da5282640,c1af91c7-6e68-40ce-b869-02fc81afa51e' \
  -H 'User-Agent: PostmanRuntime/7.20.1' \
  -H 'cache-control: no-cache' \
  -H 'content-type: multipart/form-data; boundary=----WebKitFormBoundary7MA4YWxkTrZu0gW' \
  -F file=@/home/alepox/Desktop/tp_prof/rpl/pruebita_runner/la_submission.tar.xz \
  -F 'description=files for C'

Just the consumer (without using rabbitmq)

If you already have submissions in the DB and the RPL-Server is up and running.

python3 receiver.py <submission_id> <lang>

Just the consumer (using rabbitmq)

If you already have submissions in the DB and the RPL-Server is up and running and the rabbitmq server too.

Enqueue a submissionId and the magic will do the rest

python3 rabbitmq_receive.py

python3 rabbitmq_send.py <submission_id> <lang>

Compiling and deleting source files

Check out c_Makefile and python_Makefile

C

We are just compiling the code and deleting the source files before execution

python

First we execute python -m py_compile so that, if it fails, we get a nice "user-only-code" error message We are using (pyinstaller)[https://pyinstaller.readthedocs.io/en/stable/usage.html] to generate a binary file and then executing it in the "run" step.