Test proxy #17
Test proxy #17
Conversation
tests/test_smart_proxy.py
Outdated
| @pytest.fixture | ||
| def content_smart_proxy(api, organization): | ||
| content_proxy_id = '2' | ||
| content_proxy = api.resource('smart_proxies').call('show', {'id': content_proxy_id}) |
There was a problem hiding this comment.
@ekohl So I looked at picking the content proxy from the API... but the mirror property did not appear to be a capability and rather was a setting (which I didnt see in the API response).
There was a problem hiding this comment.
Ah right, we don't expose settings since they may contain secrets. For example, it can expose a username/password which we don't want to leak. I think we need a way to declare which settings we do expose (because they are safe).
Edit: can't we list here and check that it at least is a valid ID? You can probably list all the ones with the pulpcore feature and take the one with the highest ID.
There was a problem hiding this comment.
Can we add mirror into the set of capabilities? I tried looking at the smart-proxy code but was not as clear to me where to do that so it got the right value.
There was a problem hiding this comment.
ehelms
force-pushed
the
test-proxy
branch
2 times, most recently
from
1bc6b61 to
9119339
Compare
| content_proxy = proxy | ||
| break | ||
|
|
||
| content_proxy_id = content_proxy['id'] |
There was a problem hiding this comment.
If the proxy is not found, this raises a KeyError. You will hit this guaranteed on a Pulp 3 only setup where you need to search for a Pulpcore feature with the mirror setting set to true. I don't think settings are exposed (the are considered private) so we should figure out a different way to detect this. Perhaps it should be a capability (which I think is exposed).
| content_proxy_id = content_proxy['id'] | ||
|
|
||
| environments = api.resource('capsule_content').call('lifecycle_environments', { | ||
| 'id': content_proxy['id'] |
There was a problem hiding this comment.
Any reason you don't use content_proxy_id here?
| 'id': content_proxy['id'] | |
| 'id': content_proxy_id |
| return api.resource('organizations').call('index', { | ||
| 'search': entities['organization_label'] | ||
| })['results'][0] |
There was a problem hiding this comment.
I wonder if we can use get an org by label (show action) but https://theforeman.org/api/2.3/apidoc/v2/organizations/show.html doesn't really clarify this. I know in some places we can use some name instead of just the DB ID.
| })['results'][0] | ||
|
|
||
|
|
||
| def test_smart_proxy_content_sync(api, content_smart_proxy): |
There was a problem hiding this comment.
This test will fail on a vanilla Foreman where Pulp is not present. That's something we need to deal with in a generic way. We'll see similar things with other plugins.
If we take REX: if present, it should be tested. If it isn't, it should be skipped.
That does leave the situation where a plugin should be present. If it's skipped, we need a way to fail in case it's missing. If we take the plugins testing pipeline, we should have some way to say "expect at least these plugins to be present or fail".
Working to replace https://github.com/theforeman/forklift/blob/master/bats/fb-katello-proxy.bats which I believe is a two part test and I would like to split it up. The first part is testing that a content proxy sync, and the second is that content is available which feels more like testing a client against a content proxy. The latter part is a set of additional testing that I would like to add after initial client testing goes into the code base.