trouble running atomicapp with oc new-app on openshift

[jasonbrooks]

I've running origin 1.3 in docker containers on centos atomic host continuous. Maybe I'm doing this wrong? I've tried w/ a few different atomicapps, with the same result. I got a similar error last week with a oc cluster up install, and I just got it on adb, too. All the same error.

Here's what I'm running:

# oc new-app projectatomic/guestbookgo-atomicapp --grant-install-rights                                                                                             
--> Installing application from "projectatomic/guestbookgo-atomicapp"
    * Found Docker image 510115a (9 days old) from Docker Hub for "projectatomic/guestbookgo-atomicapp"
    * Install will run in pod guestbookgo-atomicapp-install
    * The pod has access to your current session token via environment variable TOKEN_ENV_VAR.
      If you cancel the install, you should delete the pod or log out of your session.
--> WARNING: The pod requires access to your current session token to install this image. Only
      grant access to images whose source you trust. The image will be able to perform any
      action you can take on the cluster.

This is the log:

/usr/lib/python2.7/site-packages/urllib3/connectionpool.py:769: InsecureRequestWarning: Unverified HTTPS request is being made. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.org/en/latest/security.html
  InsecureRequestWarning)
DEBUG  :: - utils.py :: Detected environment: openshift pod
INFO   :: - cli/main.py :: Atomic App: 0.6.3 - Mode: Run
DEBUG  :: - cli/main.py :: Final parsed cmdline: run -v --dest=none --provider=openshift /application-entity
DEBUG  :: - nulecule/main.py :: 'none' destination requested. Using tmp dir
DEBUG  :: - nulecule/main.py :: NuleculeManager init app_path: /tmp/atomicappabHTkc
DEBUG  :: - nulecule/main.py :: NuleculeManager init image: None
DEBUG  :: - nulecule/main.py :: Request to unpack to None to /tmp/atomicappabHTkc
INFO   :: - nulecule/base.py :: Pulling external application: redis-centos7-atomicapp
INFO   :: - nulecule/base.py :: Unpacking image projectatomic/redis-centos7-atomicapp to /tmp/atomicappabHTkc/external/redis-centos7-atomicapp
DEBUG  :: - utils.py :: Detected environment: openshift pod
DEBUG  :: - plugin.py :: Configuration option 'provider-config' not provided
DEBUG  :: - providers/openshift.py :: Given config: {'namespace': 'default'}
INFO   :: - providers/openshift.py :: Using namespace default
DEBUG  :: - utils.py :: Finding the users home directory
DEBUG  :: - utils.py :: Running as user root. Using home directory /root for configuration data
DEBUG  :: - providers/openshift.py :: List of parameters passed: {'provider-tlsverify': None, 'provider-api': None, 'provider-cafile': None, 'provider-auth': None}
DEBUG  :: - providers/openshift.py :: List of parameters passed: {'provider-tlsverify': None, 'provider-api': 'http://localhost:8443', 'provider-cafile': None, 'provider-auth': None}
ERROR  :: - cli/main.py :: You need to set provider-auth in answers.conf or pass it as a CLI param

[dustymabe]

looks like the "detection" that atomicapp is running on openshift and the special behavior that results was affected by 1a37ab8

after that commt the self.access_token = os.environ['TOKEN_ENV_VAR'] is no longer done like was done here:

atomicapp/atomicapp/providers/openshift.py

Lines 629 to 636 in 4f75612

	if Utils.running_on_openshift():
	self.providerapi = Utils.get_openshift_api_endpoint_from_env()
	self.namespace = os.environ['POD_NAMESPACE']
	self.access_token = os.environ['TOKEN_ENV_VAR']
	self.provider_ca = OPENSHIFT_POD_CA_FILE
	self.provider_tls_verify = \
	self.config.get(PROVIDER_TLS_VERIFY_KEY, True)
	return # No need to process other information

so yes, confirmed that this is broken

[cdrage]

@dustymabe @jasonbrooks once I have some time to devote to atomicapp I can have a look at fixing this.