Skip to content
This repository has been archived by the owner on Apr 22, 2023. It is now read-only.

test,win: disable test-tls-server-verify for CI #25284

Closed
wants to merge 1 commit into from
Closed

test,win: disable test-tls-server-verify for CI #25284

wants to merge 1 commit into from

Conversation

orangemocha
Copy link
Contributor

test-tls-server-verify takes a lont time to execute and times
out on the Jenkins machines.

test-tls-server-verify takes a lont time to execute and times
out on the Jenkins machines.
@orangemocha
Copy link
Contributor Author

cc @joyent/node-coreteam : let's disable this test to unblock the CI. I am investigating the issue with P-1: #25279

@jasnell
Copy link
Member

jasnell commented May 12, 2015

+1
On May 12, 2015 8:51 AM, "orangemocha" notifications@github.com wrote:

cc @joyent/node-coreteam
https://github.com/orgs/joyent/teams/node-coreteam : let's disable this
test to unblock the CI. I am investigating the issue with P-1: #25279
#25279


Reply to this email directly or view it on GitHub
#25284 (comment).

@indutny
Copy link
Member

indutny commented May 12, 2015

-1 from me. This test is passing like a charm on io.js.

@orangemocha
Copy link
Contributor Author

@indutny this PR doesn't actually disable the test in general, it just lets node-accept-pull-request ignore its failures.

@orangemocha
Copy link
Contributor Author

..that is, you need to pass --flaky-tests=dontcare to the test runner to see any difference. We only do that in node-accept-pull-request

@indutny
Copy link
Member

indutny commented May 12, 2015

It is kind of the similar to disabling it, right? ;) what's the point if the test should be run by CI anyway?

@orangemocha
Copy link
Contributor Author

it's only disabling it in the context of the automated merge job. The alternative is to block accepting all other pull requests until this test is fixed.

@orangemocha
Copy link
Contributor Author

btw, the same problem seem to affect io.js: nodejs/node#1461. This test is slow only on Windows.

orangemocha added a commit that referenced this pull request May 12, 2015
test-tls-server-verify takes a lont time to execute and times
out on the Jenkins machines.

Reviewed-By: James M Snell <jasnell@gmail.com>
PR-URL: #25284
@indutny
Copy link
Member

indutny commented May 12, 2015

@orangemocha this is absolutely correct, but nevertheless it does not seem to be a reason to disable it, at least for me.

@orangemocha
Copy link
Contributor Author

@indutny does that mean you think we should stop accepting pull requests until this test is fixed? It has been slow on Windows forever, it just so happens that now it's crossing the 60 seconds mark.

@jasnell
Copy link
Member

jasnell commented May 12, 2015

If the test is timing out because of a known system specific issue but we can verify that the test actually does pass independent of the automated run, then there should be no problem here. Omitting it from the automated pass should not be a problem. However, there should be a plan to address the issue and get the test re-enabled so we need to make sure that this is revisited.

@Fishrock123
Copy link
Member

So just a note: io.js prefers to mentally "ignore" errors rather than have a "green" CI run with ignored tests.

(I.e. having a not-CI-green build is OK in these instances and more transparent.)

I have a feeling this error was fixed on our end by a recent jenkins upgrade, although I may be mistaken.

@jasnell
Copy link
Member

jasnell commented May 12, 2015

It's entirely possible. I've seen these kinds of timeout errors magically disappear across jenkins versions before. In this case, the jenkins job is not able to simple ignore the error that occurs.

@misterdjules
Copy link

@jasnell @Fishrock123 See my comment:

I did some additional tests, and test/simple/test-tls-server-verify.js takes around 80 seconds to run on our Windows 2012 server Jenkins agent.

@Fishrock123
Copy link
Member

Right, I'm not saying this isn't slow, but I think the entire CI-must-be-100%-"green" / not-"blocked" mantra may be unsavoury / detrimental.

(Admittedly, finding actual errors is easier when jenkins isn't trolling us, but still.)

@misterdjules
Copy link

I just wanted to clarify that the issue is probably not due to Jenkins, since the test takes 80 seconds to complete when I log in with remote desktop on the machine and I run this single test manually.

@misterdjules
Copy link

@Fishrick123 Yes, I agree that it's still not entirely clear what is more practical. However I think what @jasnell said previously would work well until this issue is fixed:

If the test is timing out because of a known system specific issue but we can verify that the test actually
does pass independent of the automated run, then there should be no problem here. Omitting it from
the automated pass should not be a problem. However, there should be a plan to address the issue
and get the test re-enabled so we need to make sure that this is revisited.

With an emphasis on: but we can verify that the test actually does pass independent of the automated run.

@orangemocha
Copy link
Contributor Author

Right, I'm not saying this isn't slow, but I think the entire CI-must-be-100%-"green" / not-"blocked" mantra may be unsavoury / detrimental.

@Fishrock123 , the point is not for the entire CI to be green. All other CI jobs except for node-accept-pull-request will show red when this test fails. Even node-accept-pull-request will not be green. It will be yellow, which gives us a clue that not everything is ok.

The test is marked as "not ok", but there's a TODO next to it, which tells node-accept-pull-request to continue with the merge.
See here for an example http://jenkins.nodejs.org/job/node-test-commit-windows/74/DESTCPU=ia32,label=windows/tapTestReport/

@orangemocha
Copy link
Contributor Author

Anyway, this PR already landed after James' +1 because I thought it was urgent to unblock the CI. Since it ended up being a bit controversial, we can discuss it at the next core team meeting, and decide then if it should stay or be reverted.

@orangemocha
Copy link
Contributor Author

Landed in 0dbaee4

@misterdjules
Copy link

Thank you @orangemocha 👍 I think that with some extra care until the test is fixed, this is the best solution for now.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

6 participants