Locale issue on fresh install of Fuel 7.0

[Sebastian Kalinowski]

 When calling fuel2 and some error occur (like invalid command name) traceback is printed:

[root@nailgun ~]# fuel2 invalid-command-name
DEPRECATION WARNING: /etc/fuel/client/config.yaml exists and will be used as the source for settings. This behavior is deprecated. Please specify the path to your custom settings file in the FUELCLIENT_CUSTOM_SETTINGS environment variable.
Traceback (most recent call last):
  File "/usr/bin/fuel2", line 10, in <module>
    sys.exit(main())
  File "/usr/lib/python2.6/site-packages/fuelclient/cli/error.py", line 98, in wrapper
    return func(*args, **kwargs)
  File "/usr/lib/python2.6/site-packages/fuelclient/main.py", line 75, in main
    command_manager=CommandManager('fuelclient', convert_underscores=True)
  File "/usr/lib/python2.6/site-packages/cliff/app.py", line 73, in __init__
    self._set_streams(stdin, stdout, stderr)
  File "/usr/lib/python2.6/site-packages/cliff/app.py", line 80, in _set_streams
    locale.setlocale(locale.LC_ALL, '')
  File "/usr/lib64/python2.6/locale.py", line 513, in setlocale
    return _setlocale(category, locale)
locale.Error: unsupported locale setting

After setting LC_ALL variable it works again.

custom ISO:

VERSION:
  feature_groups:
    - mirantis
  production: "docker"
  release: "7.0"
  openstack_version: "2015.1.0-7.0"
  api: "1.0"
  build_number: "891"
  build_id: "2015-08-10_11-42-06"
  nailgun_sha: "ee00c45985ed05a4b246ae5966fb7e8f44f1095d"
  python-fuelclient_sha: "e965951d94014920831499197e07a0e1a7c52f37"
  fuel-agent_sha: "57145b1d8804389304cd04322ba0fb3dc9d30327"
  fuel-nailgun-agent_sha: "e01693992d7a0304d926b922b43f3b747c35964c"
  astute_sha: "e1d3a435e5df5b40cbfb1a3acf80b4176d15a2dc"
  fuel-library_sha: "38a188ebf4142744edf9c00e99dc9d97666bd724"
  fuel-ostf_sha: "c7f745431aa3c147f2491c865e029e0ffea91c47"
  fuelmain_sha: "b87de67f15317bea2fa826d0b367e6f29b798813"

[Roman Prykhodchenko (romcheg)]

Apart from proposing a patch to Cliff it's also a must to set LC_ALL for root and in /etc/skell: that will resolve the issue faster. Unset LC_ALL will cause failures not only in Fuel Client.

[Bartłomiej Piotrowski (bpiotrowski)]

Still, cliff should try to fall back to C locales if LC_* is unset.

[Alexander Bochkarev (abochkarev)]

I've just found the bug https://bugs.launchpad.net/fuel/+bug/1458361 that describes the same behavior. From the discussion it became clear that this is known-issue which doesn't affect deployment process. Was decided to move it to 8.0.

[Alexander Bochkarev (abochkarev)]

Forgot to add.

VERSION:
  feature_groups:
    - mirantis
  production: "docker"
  release: "7.0"
  openstack_version: "2015.1.0-7.0"
  api: "1.0"
  build_number: "164"
  build_id: "2015-08-12_15-16-47"
  nailgun_sha: "fff6bda090fac15c48b27cca7832a70f8e381101"
  python-fuelclient_sha: "26fc025e0fc5791b62e5ed8561a6016bf8a406bc"
  fuel-agent_sha: "57145b1d8804389304cd04322ba0fb3dc9d30327"
  fuel-nailgun-agent_sha: "e01693992d7a0304d926b922b43f3b747c35964c"
  astute_sha: "e1d3a435e5df5b40cbfb1a3acf80b4176d15a2dc"
  fuel-library_sha: "22f848670e49d89fc04aaed4d8efd1b07360cbe7"
  fuel-ostf_sha: "58220583f10fa47f12291488ef77854809c68310"
  fuelmain_sha: "67e5214c0dc5d4ba6da4ae651cef9934800459a9"

[root@fuel ~]# fuel2 invaild-command-name
DEPRECATION WARNING: /etc/fuel/client/config.yaml exists and will be used as the source for settings. This behavior is deprecated. Please specify the path to your custom settings file in the FUELCLIENT_CUSTOM_SETTINGS environment variable.
Unknown command ['invalid-command-name']

There are no errors. Only deprecation warning.

[Ihor Kalnytskyi (ikalnytskyi)]

The problem's solved. Sebastian has wrong LC_ALL setup on his system -

LANG="en_US.UTF-8"
LC_COLLATE="en_US.UTF-8"
LC_CTYPE="UTF-8"
LC_MESSAGES="en_US.UTF-8"
LC_MONETARY="en_US.UTF-8"
LC_NUMERIC="en_US.UTF-8"
LC_TIME="en_US.UTF-8"
LC_ALL=

and ssh by default sends LC_* settings to the remote server -

$ cat /etc/ssh/ssh_config | grep LC
   SendEnv LANG LC_*

move it it Invalid.

[Sebastian Kalinowski (prmtl)]

It's because missing LC_ALL on the host from the host we are connecting to the master node (I know it thansk to Igor Kalnitsky ;)). and when ssh connection is sending this empty LC_ALL to session on our mater node.

We can safely decrease priority and move to 8.0