Cannot retrieve test results when running `apb test`
psturc opened this issue · 9 comments
Hi, I can't get over the issue with gathering test results when running apb test
Log from pod
| TASK [Run the verify role.] ****************************************************
|
| TASK [verify-test-apb : url check for webpage status] **************************
| FAILED - RETRYING: url check for webpage status (10 retries left).
| ok: [localhost]
|
| TASK [verify-test-apb : Save failure for the web page] *************************
| skipping: [localhost]
|
| TASK [verify-test-apb : fail] **************************************************
| skipping: [localhost]
|
| TASK [verify-test-apb : Save test pass] ****************************************
| changed: [localhost]
|
| PLAY RECAP *********************************************************************
| localhost : ok=6 changed=4 unreachable=0 failed=0
|
| + EXIT_CODE=0
| + set +ex
| + '[' -f /var/tmp/test-result ']'
| + test-retrieval-init
| Waiting for the user to gather the test resutls...
| Waiting for the user to gather the test resutls...
| Waiting for the user to gather the test resutls...
The pod just ends after 5 minutes (as defined in /usr/bin/test-retrieval-init), because the test results can't be gathered from here
Log from terminal
Building APB using tag: [test-apb]
Successfully built APB image: test-apb
Creating project apb-test-test-apb
Created project
Creating service account in apb-test-test-apb
Created service account
Creating role binding for apb-test-test-apbvl4vn in apb-test-test-apb
Created Role Binding
Creating pod with image test-apb in apb-test-test-apb
Created Pod
Unable to retrieve test result.
Deleting project apb-test-test-apb
Project deleted
When I printed out the ApiException, it says
(0)
Reason: Handshake status 400
How to reproduce this:
- Clone this repo
- Edit Makefile and run
make - Run
apb test
Environment
macOS High Sierra 10.13.1
oc v3.7.0+7ed6862
kubernetes v1.7.6+a08f5eeb62
features: Basic-Auth
Server https://192.168.37.1:8443
openshift v3.7.0+7ed6862
kubernetes v1.7.6+a08f5eeb62
@shawn-hurley do you have some thoughts on this?
I am seeing the same behavior. I am wondering if a new version of the python client is needed. I think that the python client is attempting to hit an old pod execution endpoint. I will dig into this and try and get a fix in. @psturc I will keep you updated on this bug as I dig into it.
@fabianvf @djzager I might need your help.
So I found that it appears according to the swagger spec it appears that the Method that is expected is a GET request. The python kubernetes client v3.0 appears to attempt a post request at line 4059. I think that we need to resolve this mismatch for this to be able to connect to the running pod and complete the job.
@psturc I was testing with our environment and was seeing the same thing you were. I am going to try and see if this is the same issue that you are seeing.
I have also verified that this would occur with the version of k8s that you are using with openshift.
@shawn-hurley the method you linked to from the swagger spec maps to connect_get_namespaced_pod_exec, there are other methods on that path with corresponding methods in the client-python's core_v1_api but it isn't clear to me yet what needs to be reconciled.
The swagger spec only has a endpoint that accepts get methods.
The client python has a get and post, currently, the apb tool is using the connect_post_namespaced_pod_exec. Because the swagger spec does not have a defined endpoint that I can see for a POST request I think this is why the apb tool cannot connect to the pod.
@shawn-hurley is this not the POST endpoint? https://github.com/kubernetes/kubernetes/blob/release-1.8/api/swagger-spec/v1.json#L10241
I think you nad @djzager are correct. I think I am wrong. This might be some other error then.