Local ansible deployment API gateway error 502

[nslvn]

Environment details:

• local deployment, native ubuntu, Mac OS, Kubernetes, ...
  Local deployment using ansible with apigateway
• version of docker, ubuntu, ...
  Docker version 24.0.7, build afdd53b
  Ubuntu 22.04.3 LTS

Steps to reproduce the issue:

1. Deploy openwhisk using ansible

export ENVIRONMENT=local
# Export Keys
ansible-playbook -i environments/$ENVIRONMENT prereq.yml
ansible-playbook -i environments/$ENVIRONMENT setup.yml
# couchdb is still needed to store subjects and actions
ansible-playbook -i environments/$ENVIRONMENT couchdb.yml
ansible-playbook -i environments/$ENVIRONMENT initdb.yml
ansible-playbook -i environments/$ENVIRONMENT wipe.yml

ansible-playbook -i environments/$ENVIRONMENT openwhisk.yml \
-e skip_pull_runtimes=True \
-e limit_action_memory_max=4294967296 \
-e limit_action_memory_std=2147483648 \
-e limit_invocations_per_minute=999999 \
-e limit_invocations_concurrent=9999999 \
-e limit_fires_per_minute=9999999 \
-e limit_sequence_max_length=1000 \
-e namespace_default_limit_action_memory_max=8192 \
-e lean=true \
-e invoker_user_memory="16384m" # Too much here is an issue when it creates too many concurrent processes!

# to use the API gateway
ansible-playbook -i environments/$ENVIRONMENT apigateway.yml
ansible-playbook -i environments/$ENVIRONMENT routemgmt.yml

ansible-playbook -i environments/$ENVIRONMENT edge.yml \
    -e cli_installation_mode=remote \
    -e openwhisk_cli_home="$OPENWHISK_HOME/../openwhisk-cli"

# installs a catalog of public packages and actions
ansible-playbook -i environments/$ENVIRONMENT postdeploy.yml

2. Create an action wsk -i action create test test.py
3. Create an API route wsk -i api create /test get /guest/test
4. Curl the action -> Error 502

Provide the expected results and outputs:

A working response (not 502)

Provide the actual results and outputs:

{"code":502, "message":"Oops. Something went wrong. Check your URI and try again."}

Additional information you deem important:

I've found this issue (apache/openwhisk-apigateway#370) for the apigateway already. However, I think the bug is with the deployment script, not the gateway itself. Adding "BACKEND_HOST" in the apigateway task deploy role file fixes this issue, but I'm not sure if this is correct for other deployments as well. Note that using "http" as a scheme (as seems to be used by the standalone version) does not work because the proxy does forward the 308 redirect to https instead of handling it in-line.

- name: (re)start apigateway
  docker_container:
    name: apigateway
    image: "{{ apigateway.docker_image | default('openwhisk/apigateway:' ~ apigateway.version) }}"
    state: started
    recreate: true
    restart_policy: "{{ docker.restart.policy }}"
    hostname: apigateway
    env:
      "REDIS_HOST": "{{ groups['redis'] | first }}"
      "REDIS_PORT": "{{ redis.port }}"
      "REDIS_PASS": "{{ redis.password }}"
      "PUBLIC_MANAGEDURL_HOST": "{{ ansible_host }}"
      "BACKEND_HOST": "https://{{ groups['controllers']  | map('extract', hostvars, 'ansible_host') | first }}" # Added to avoid gateway issues
      "PUBLIC_MANAGEDURL_PORT": "{{ apigateway.port.mgmt }}"
      "TZ": "{{ docker.timezone }}"
    ports:
      - "{{ apigateway.port.mgmt }}:8080"
      - "{{ apigateway.port.api }}:9000"
    pull: "{{ apigateway_local_build is undefined }}"

- name: wait until the API Gateway in this host is up and running
  uri:
    url: "http://{{ groups['apigateway'] | first }}:{{ apigateway.port.api }}/v1/apis"
  register: result
  until: result.status == 200
  retries: 12
  delay: 5