Handle wrong provider name in answers.conf

[surajssd]

If the provider name for the atomicapp in answers.conf is mistaken(other than docker, kubernetes, openshift, marathon) then it fails giving all the stack trace without any sane error. Re-creating the errorsome condition:

(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$ pwd
/home/vagrant/nulecule-library/apache-centos7-atomicapp
(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$ tree
.
├── answers.conf
├── answers.conf.orig
├── artifacts
│   ├── docker
│   │   └── apache-centos7-atomicapp_run
│   ├── kubernetes
│   │   └── apache-centos7-atomicapp-pod.json
│   └── marathon
│       └── apache-centos7-atomicapp.json
├── Dockerfile
├── Nulecule
└── README.md

4 directories, 8 files
(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$ cat answers.conf
[apache-centos7-atomicapp]
image = centos/httpd
hostport = 80
[general]
namespace = default
provider = something
(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$ sudo atomicapp run . -v
2016-02-16 07:00:14,739 - atomicapp.cli.main - INFO - Action/Mode Selected is: run
2016-02-16 07:00:14,739 - atomicapp.cli.main - DEBUG - Final parsed cmdline: run . -v
2016-02-16 07:00:14,739 - atomicapp.nulecule.main - DEBUG - NuleculeManager init app_path: .
2016-02-16 07:00:14,739 - atomicapp.nulecule.main - DEBUG - NuleculeManager init    image: None
2016-02-16 07:00:14,739 - atomicapp.utils - DEBUG - Loading answers from file: ./answers.conf
2016-02-16 07:00:14,740 - atomicapp.nulecule.main - DEBUG - Request to unpack to None to .
2016-02-16 07:00:14,744 - atomicapp.plugin - DEBUG - Loading providers from /home/vagrant/atomicapp/atomicapp/providers
2016-02-16 07:00:14,744 - atomicapp.plugin - DEBUG - Loading providers from /home/vagrant/atomicapp/atomicapp/providers
2016-02-16 07:00:14,746 - atomicapp.cli.main - ERROR - 'NoneType' object is not callable
Traceback (most recent call last):
  File "/home/vagrant/atomicapp/atomicapp/cli/main.py", line 100, in cli_run
    nm.run(**argdict)
  File "/home/vagrant/atomicapp/atomicapp/nulecule/main.py", line 228, in run
    self.nulecule.run(cli_provider, dryrun)
  File "/home/vagrant/atomicapp/atomicapp/nulecule/base.py", line 148, in run
    provider_key, provider = self.get_provider(provider_key, dryrun)
  File "/home/vagrant/atomicapp/atomicapp/nulecule/lib.py", line 99, in get_provider
    self.get_context(), self.basepath, dry)
TypeError: 'NoneType' object is not callable
(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$

But when --provider is given from command line then it works perfectly fine:

(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$ sudo atomicapp run . --provider=kubrnetes
usage: atomicapp run [-h] [-V] [-v] [-q] [--mode {fetch,run,stop,genanswers}]
                     [--dry-run] [--answers-format {ini,json,xml,yaml}]
                     [--namespace NAMESPACE]
                     [--providertlsverify {True,False}]
                     [--providerconfig PROVIDERCONFIG]
                     [--providercafile PROVIDERCAFILE]
                     [--providerapi PROVIDERAPI] [-a ANSWERS]
                     [--write-answers ANSWERS_OUTPUT]
                     [--provider {docker,kubernetes,openshift,marathon}]
                     [--ask] [--destination DESTINATION]
                     [app_spec]
atomicapp run: error: argument --provider: invalid choice: 'kubrnetes' (choose from 'docker', 'kubernetes', 'openshift', 'marathon')
(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$ sudo atomicapp run . --provider=something
usage: atomicapp run [-h] [-V] [-v] [-q] [--mode {fetch,run,stop,genanswers}]
                     [--dry-run] [--answers-format {ini,json,xml,yaml}]
                     [--namespace NAMESPACE]
                     [--providertlsverify {True,False}]
                     [--providerconfig PROVIDERCONFIG]
                     [--providercafile PROVIDERCAFILE]
                     [--providerapi PROVIDERAPI] [-a ANSWERS]
                     [--write-answers ANSWERS_OUTPUT]
                     [--provider {docker,kubernetes,openshift,marathon}]
                     [--ask] [--destination DESTINATION]
                     [app_spec]
atomicapp run: error: argument --provider: invalid choice: 'something' (choose from 'docker', 'kubernetes', 'openshift', 'marathon')
(atomicapp) [vagrant@centos7-adb apache-centos7-atomicapp]$

The fix is, to raise an error if given provider_key matches none of the existing key in getProvider method of class Plugin in plugin.py. Shortcut to function here.

[surajssd]

Working on the fix.

[surajssd]

@dustymabe @jberkus PR for this issue is merged, so this remains open ?

[dustymabe]

Looks good:

[root@f23 helloapache (dusty-fix-mariadb *%)]# aa run ./ 
2016-03-09 16:45:32,138 - [INFO] - main.py - Action/Mode Selected is: run
2016-03-09 16:45:32,157 - [ERROR] - main.py - Invalid Provider - 'kubernetess', provided in answers.conf (choose from docker, kubernetes, openshift, marathon)