Kubernetes load_config failure while executing from within the POD

[vrock28]

Here is the code:

from kubernetes import config
import kubernetes.client
from kubernetes.client.rest import ApiException

config.load_incluster_config()
api_instance = kubernetes.client.CoreV1Api()
...
.........

Error:

config.load_incluster_config()\n File "local/lib/python2.7/site-packages/kubernetes/config/incluster_config.py",
line 93, in load_incluster_config\n cert_filename=SERVICE_CERT_FILENAME).load_and_set()\n
File "local/lib/python2.7/site-packages/kubernetes/config/incluster_config.py",
line 45, in load_and_set\n
self._load_config()\n
File "local/lib/python2.7/site-packages/kubernetes/config/incluster_config.py",
line 51, in _load_config\n
raise ConfigException("Service host/port is not set.")\n
ConfigException: Service host/port is not set.'

python --version

Python 2.7.6

python Kubernetes client version 4.0.0 is installed

When i try to execute the statements directly inside the POD,
i see that neither the host or port info is missing
in the environment.

python
Python 2.7.6 (default, Oct 26 2016, 20:30:19)
[GCC 4.8.4] on linux2
Type "help", "copyright", "credits" or "license" for more information.

import os

print os.environ["KUBERNETES_SERVICE_HOST"]
10.3.0.1
print os.environ["KUBERNETES_SERVICE_PORT"]
443

[1]+ Stopped python

[kevin-bates]

@vrock28 - not sure if this is applicable, but I ran into this issue yesterday. I have a pod (service) that spawns other pods (jobs), then monitors/controls them via the python api. The monitoring calls were working, but the job-creation calls were producing this error.

The issue was because I use a sub-process (via Popen()) from which the jobs are created, yet didn't "flow" the service host and port envs to the sub-process. Once those two values were propagated, load_incluster_config() could complete and the jobs were appropriately created.

[vrock28]

@kevin-bates - Thanks for the pointer. My POD runs a Flask application and i am not sure if it actually spawns a sub-process in some way. Nonetheless, how did you manage to prppogate the values to the sub-process ?

[kevin-bates]

@vrock28 - Popen() takes a **kwargs, one of which is an env dictionary. As a result, I could do the following:

    # Transfer internal env vars since the launcher starts fresh
    kw['env']['KUBERNETES_SERVICE_HOST'] = os.environ.get('KUBERNETES_SERVICE_HOST')
    kw['env']['KUBERNETES_SERVICE_PORT'] = os.environ.get('KUBERNETES_SERVICE_PORT')

[DrJimFan]

@vrock28 @kevin-bates I'm experiencing exactly the same error with the same code. I'm running the python client outside any pods or subprocesses, so env variable propagation is likely not the cause.

[DrJimFan]

Are you able to resolve this problem? @vrock28

[vrock28]

@linxifan - No i am still not able to resolve this error. I hacked a bit to get around the issue by hard coding the HOST and PORT

[DrJimFan]

@vrock28 I'm using python 3.6, btw, so it's not your python's problem.

[DrJimFan]

@vrock28 even after I hardcode the HOST and PORT, I'm still getting "Service token not found" error. Something in the env is seriously wrong ...

[vrock28]

@linxifan - I just edited the incluster_config.py and removed the code which checks for the env variables.

[vrock28]

@linxifan - May be trying with earlier rlse of kubernetes ( < 4.0.0) would help ?

[DrJimFan]

@vrock28 I'm giving up. For my use case, I can simply use subprocess module to call kubectl directly and process the stdout. The API is poorly documented and not worth the trouble (see #437 ).

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle rotten