-
Notifications
You must be signed in to change notification settings - Fork 1.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove overlapping tests with Prow #1350
Remove overlapping tests with Prow #1350
Conversation
Hi @Adirio. Thanks for your PR. I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
This comment has been minimized.
This comment has been minimized.
Just a comment about this part of your proposal. E2E tests can't be performed in OSX. About the issue you comment: not being able to see the result of e2e tests before So I think the question we should debate is, which tests should we gate behind the Some things to bear in mind (correct me if I'm wrong):
My opinion:
If it wasn't for the |
This comment has been minimized.
This comment has been minimized.
With your proposal, which is the current behavior, I don't think the costs as you said are a real issue, as the curren trend has been towards using prow. The |
This comment has been minimized.
This comment has been minimized.
That's basically nothing. Most of the tests would already be run without an aproval of a reviewer.
Not protecting something from an attack just because a certain user can manually block the attack is a recipe for a disaster in ciber-security.
Not true, if you open this PRs Travis, you will see that it uses the Travis config from this PR, instead of using it from master. That may be another thing we may want to change, but in that case we would not know if a change to |
This comment has been minimized.
This comment has been minimized.
Hi @Adirio, Also, I forgot to share that Travis will kill a process that did not finish after X time as it has a lot of mechanisms to prevent scenarios as you described besides all be executed in isolate envs. So, you do not need to have concerns such as over an attack against it or if it is safe to be used. It is a commonly used and very mature project. |
Unitary tests can (and should) be run locally before PR as part of the development process. Coverage also executes unitary tests. Linting is also performed. The prupose of
I'm not saying that they are going to break Travis, but they can overload it. DoS attack. Good PRs will not have resources to pass their Travis tests because Travis is busy with rogue PRs.
Travis should obviouslly use the code from PR, but we can probably enforce it to use the |
The intent of IMO, we should try to move most things to prow, since it's easier to do stuff like manually retrigger runs & prevent folks from abusing our lint scripts. Plus, anything on prow gets run as part of merge, so the more we put there, the less likely we are to have scenarios like merges breaking lints. |
This comment has been minimized.
This comment has been minimized.
I agree with some, disagree with others.
Exactly, that is the purpose of
Why would you need to run all the tests to know if you can set a flag that allows to run tests? It doesn't make sense.
They are not always faster. Prow tests may take a bit longer if they both start at the same time, but travis processes are limited, so when there are a couple PRs, which is pretty common, all processes will not start and the waiting time can be quite big.
True, and that seems to not be something temporal. It doesn't look as supporting other OS is coming soon. So basically I would say:
*1 To make coverage part of the merge process, we should probably investigate how to tell coverall.io which PR it is coming from and all that stuff. Also, it would be needed develop a lot of tests for kubebuilder, which has almost no unit tests (which are the ones reporting coverage, e2e tests do not help with this). |
72164a4
to
e4b49e8
Compare
Signed-off-by: Adrian Orive <adrian.orive.oneca@gmail.com>
IMO contributors are supposed to ensure tests are passing locally before submitting a PR.
We instruct them in the failure message: Line 49 in 48ea4cb
This is a big advantage of Prow over Travis. Prow is smarter to figure out if a test result is considered to be out-of-date. IIRC Prow currently supports Linux and Windows, but not MacOS. kubebuilder supports Linux and MacOS. |
Windows? I was told today that prow runs over k8s and that k8s doesn't run on native windows (docker on windows is basically a linux kernel with HyperV). But the point is clear, it doesn't support MacOS and kubebuilder does. So this PR removes e2e tests from Travis as they are already in Prow, before moving the golden test we will need to implement it in Prow. |
It seems Prow allow using Windows nodes in a k8s cluster. |
I created #1357 to track the contributor doc efforts |
I am cool with the decision to move these tests to Prow.
IHMO: we need at least track this needs in an issue. WDYT? Is it make sense? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
After the discussions over the topic. I think that it is ok to be merged since it is just removing the duplication of the tests. I mean, it is just removing from Travis the tests that are already done in Prow.
So, for me it is:
/lgtm
PS.: IHMO we could/should track the points discussed in new issues. Let me know if you agree with @mengqiy and @Adirio and if has something that I can do to help.
/lgtm |
/approve |
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: Adirio, mengqiy The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
It needs a decent amount of work to support it in windows. |
Description
Prow and Travis are overlapping. This PR removes the e2e tests from Travis but leaves the local tests, as the OSX version isn't run by prow.
Motivation
#1344