openshift / openshift-client-python Goto Github PK

Please add the ability of executing: 'image mirror'

Context() class allows to override many options that would be passed as cmdline arguments, like token or server url, but I cannot find a way how to override --context cmdline option to enable having one kubeconfig and referencing different contexts within.

Is there a way how to do this currently?

It has come to our attention that the "openshift" namespace is being used by another Openshift python project:
openshift -> openshift-restclient-python

Each project provides their own version of init.py, each with conflicting setup/configuration. Because of this, these 2 projects can not happily co-exist together in the same environment

Opening this issue to re-home our package to allow for these packages to co-exist.

The following snippets have an issue because the export verb in the oc utility does not include the flag --ignore-not-found. A program will fail because objects(exportable=True) will silently pass True for the ignore flag and error out the CLI tool.

A catch as simple as adjusting L374 to if ignore_not_found and not exportable: should suffice (though it'd probably be a good idea to log or print that that flag is being ignored). In the long run (if this tool does see future developments), it would be good to see the export verb removed in favor of an --export argument to pass to get.

Is there any way to do 'oc run' with this project?

The 'oc exec' command supports passing a stateful set and a ready pod from the stateful set will automatically be chosen.
E.g.

oc exec -it statefulset/my-stateful-set -c foo -- env - echo

The openshift-python-client doesn't appear to support this. If you pass a statefulset to the command the execute command assumes it's a pod and tries to exec into a pod with the same name as the statefulset.

oc.selector('statefulset/my-stateful-set').object().execute(['ls'], container_name='foo')

...
  openshift.model.OpenShiftPythonException: [Error running statefulset.apps/my-stateful-set exec on ls [rc=1]: Error from server (NotFound): pods "my-stateful-set" not found]
{
    "operation": "exec",
    "status": 1,
    "actions": [
        {
            "timestamp": 1649435656752,
            "elapsed_time": 0.3165004253387451,
            "success": false,
            "status": 1,
            "verb": "exec",
            "cmd": [
                "oc",
                "exec",
                "--namespace=my-project",
                "--container=foo",
                "my-stateful-set",
                "--",
                "ls"
            ],
            "out": "",
            "err": "Error from server (NotFound): pods \"my-stateful-set\" not found\n",
            "in": null,
            "references": {
                ".stack": [
                  ...
                ]
            },
            "timeout": false,
            "last_attempt": true,
            "internal": false
        }
    ]
}                                                    

Overview

It would be great if Selector.objects() and Selector.object() would allow to specifiy which class it should create. This can be useful when you have custom class on top of APIObject for helping with managing said resource.

Example

oc.selector("pods").objects(class=DeploymentConfig) ~> List[DeploymentConfig]
oc.selector("pods").object(class=DeploymentConfig) ~> DeploymentConfig
oc.selector("pods").object() ~> APIObject

I think I could implement this if you would like this to be added.

Issue

When Red Hat OpenShift Serverless is installed and serving.knative.dev/v1 Route api exists as resource in the cluster, the v1_routes.create( .. ) method raise the following exception : openshift.dynamic.exceptions.ResourceNotFoundError: No matches found for {'api_version': 'v1', 'kind': 'Route'}.

How to reproduce the issue:

1. Install Red Hat OpenShift Serverless Operator

2. Create a KnativeServing object (as the following for example:)

   # cat << EOF  | oc apply -n knative-serving -f -
   apiVersion: operator.knative.dev/v1alpha1
   kind: KnativeServing
   metadata:
     name: knative-serving
     namespace: knative-serving
   spec: {}
   EOF

3. Check if the KnativeServing Route api exists in the cluster:

   # oc api-resources --verbs=list | grep routes          
   routes                                             route.openshift.io/v1                       true         Route
   routes                            rt               serving.knative.dev/v1                      true         Route

4. Try to create an OpenShift Route using openshift-client-python library executing the following script (adjusting the <variable> ):

   import yaml
   from kubernetes import client, config
   from openshift.dynamic import DynamicClient
   
   k8s_client = config.new_client_from_config()
   dyn_client = DynamicClient(k8s_client)
   
   v1_routes = dyn_client.resources.get(api_version='v1', kind='Route')
   route = """
   kind: Route
   metadata:
     name: test-route
   spec:
     host: test-route.<host>
     port:
       targetPort: 8080-tcp
     to:
       kind: Service
       name:  <service>
       weight: 100
     wildcardPolicy: None
     tls:
       termination: edge
   """
   
   route_data = yaml.load(route)
   resp = v1_routes.create(body=route_data, namespace='default')
   
   # resp is a ResourceInstance object
   print(resp.metadata)

   The execution of the script raise the following Exception:

   Traceback (most recent call last):
     File "createroute.py", line 11, in <module>
       v1_routes = dyn_client.resources.get(api_version='v1', kind='Route')
     File "/Users/gmeghnag/Library/Python/3.8/lib/python/site-packages/openshift/dynamic/discovery.py", line 246, in get
       raise ResourceNotFoundError('No matches found for {}'.format(kwargs))
   openshift.dynamic.exceptions.ResourceNotFoundError: No matches found for {'api_version': 'v1', 'kind': 'Route'}

Potential cause

From openshift-restclient-python

def get(self, **kwargs):
        """ Same as search, but will throw an error if there are multiple or no
            results. If there are multiple results and only one is an exact match
            on api_version, that resource will be returned.
        """

Other Information

# oc get clusterversion
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.6.15    True        False         134m    Cluster version is 4.6.15

# oc get sub -A
NAMESPACE              NAME                  PACKAGE               SOURCE             CHANNEL
openshift-serverless   serverless-operator   serverless-operator   redhat-operators   stable

If you create APIObject with a model that is missing namespace it will default to the default namespace instead of whatever namespace context has.

In case users use print_log function each file contains
start string [logs:begin]xxxt-xx-xxx-01:pod/xxx-xxxxx-xxxxxxxl-d997f557d-vpf8r->pod/xxx-xxxxxxx-xxxxxxxx- f557d-vpf8r(postgresql)========
end string [logs:end]xxxxx-xx-xxxxx-01:pod/xxx-xxxxxx-xxxxxl-d997f557d-vpf8r->pod/xxx-xxxxxx-xxxxxxx-d997f557d-vpf8r(postgresql)========

This is not convenient. For example if log has json format I can load to an editor and format this log as json file. This strings confuse formatter. Commonly speaking I don see real role of this string. This string is mandatory added in utils module. Can I fix them or anybody use these strings

We would like to stream the output of the oc command as it completes work. So far, we haven't been able to find a way; we can only get the full log once the command completes. Is there something simple I'm missing? Thanks! If there is a way, could you add an example file and/or doc showing how?

Hi

So I would like to find all pods in a namespace that do not have a specific label set.
I've been looking at following piece of documentation but it doesn't seem to mention it.
https://github.com/openshift/openshift-client-python#selectors

Selectors can also select based on kind and labels.
sa_label_selector = oc.selector("sa", labels={"mylabel":"myvalue"})

When doing this using the oc command I can do following:
oc get pod --selector='!mylabel'

However I can't seem to emulate this using this library.
Can you help me?

Please help, I don't see in your documentation mentioning these common resources.

Hi,

the base args should be --replicas not --scale.

Getting the following error.

Error: unknown flag: --scale

import openshift as oc
bc = oc.selector('bc/web-devops-bc')
bc.start_build()

Error message:

  File ".\oc_test.py", line 25, in _build_oc
    bc.start_build()
  File "C:\dev\wm-cicd-builder\build\lib\site-packages\openshift\selector.py", line 425, in start_build
    r = Selector()
TypeError: __init__() missing 1 required positional argument: 'high_level_operation'

Hello,

is there a way to share a context with subprocess library, for example:

import openshift as oc
import subprocess

HOST = 'https://localhost:6443'
USER = 'ocuser'
PASSWORD = 'password'
PROJECT = 'my-proj'

with oc.api_server(HOST):
    with oc.login(USER, PASSWORD):
        with oc.project(PROJECT):
            subprocess.run('oc get pods | grep ABCD')

I would like to execute some commands directly from the subprocess library, making sure I'm in the correct server/project context

oc.invoke('cp', ('/Users/<user>/Desktop/upload.txt', '{}:/tmp'.format(pod)))

I'm using this Python OpenShift client to upload a file from my local machine to the remote client_host and everything is working as expected. However, when I attempt to invoke a cp from my local machine it shows that the invocation works:

{
    "operation": "tracking",
    "status": 0,
    "actions": [
        {
            "timestamp": 1624989569902,
            "elapsed_time": 1.038727045059204,
            "success": true,
            "status": 0,
            "verb": "cp",
            "cmd": [
                "oc",
                "cp",
                "/Users/<user>/Desktop/upload.txt",
                "<pod-name>:/tmp"
            ],
            "out": "",
            "err": "",
            "in": null,
            "references": {
                ".client_host": "root@<ip-address>"
            },
            "timeout": false,
            "last_attempt": true,
            "internal": false
        }
    ]
}

When I ls the file it's not there. Any ideas why the invoke command works but there's no sight of the file on the pod?

Hello,

is there a way to share a context with subprocess library, for example:

import openshift as oc
import subprocess

HOST = 'https://localhost:6443'
USER = 'ocuser'
PASSWORD = 'password'
PROJECT = 'my-proj'

with oc.api_server(HOST):
    with oc.login(USER, PASSWORD):
        with oc.project(PROJECT):
            subprocess.run('oc get pods | grep ABCD')

I would like to execute some commands directly from the subprocess library, making sure I'm in the correct server/project context

For example there already is https://github.com/openshift/openshift-restclient-python and I can't imagine why there is two of them and which one should i use.

Hello,
i find and Error when using the follow command:

• oc.selector('pod/mypodname').object().execute(xxx)

i have the error:
"err": "error: invalid resource name "pod/xxxx": [may not contain '/']\n",

when looking the reason:
i find an error in the openshift/apiobject.py
in the fonction : def execute(.....)

you have to change the parameter qname() to name()

have a nice day..

The until_all method does not have a timeout. Consequently, if the caller does not define a failure function, the method waits indefinitely.

The failure function approach assumes that the caller knows all of the possible issues that could happen with the object. However, this is not always the case.

Would you be open to adding a timeout parameter to the functions?