Create a job for building --without-ssl

[danbev]

I would be great to have a job that builds and tests node configuring it using the --without-ssl flag. Currently it is easy to forget to test this option and PRs are merged without us noticing.

I noticed #643 exists and perhaps that should be reopened instead. Feel free to close this if that is this the case.

[refack]

Thanks for the issues.
I'll follow up.

[refack]

So it seems @rvagg already wrote the script for this, it just wasn't turned on (probably because the test suit failed)

[richardlau]

@refack #643 (comment)

[refack]

Job is now ready, and will pass once nodejs/node#24376 land: https://ci.nodejs.org/job/refack-node-test-commit-linux-containered/15/

[danbev]

Job is now ready, and will pass once nodejs/node#24376 land: https://ci.nodejs.org/job/refack-node-test-commit-linux-containered/15/

Awesome, thanks!

[rvagg]

I honestly don't know why that was never enabled (just requires adding a label to the job iirc), perhaps it was due to too many failures without openssl.

[refack]

I can imagine it was because the test suit was failing, and in several places, not just tests, for example calls to npm don't work. But now nodejs/node#24376 should be the last bloker.

[refack]

With nodejs/node#24376 landed 3 days ago, PRs should be either fresh, or "rebasable", so IMHO we're ready to enable.
Sanity on master https://ci.nodejs.org/job/refack-node-test-commit-linux-containered/16/

[refack]

Oooof, new test fails - test.parallel/test-cli-node-print-help

[danbev]

Oooof, new test fails - [test.parallel/test-cli-node-print-help](https://ci.nodejs.org/job/refack-

I just ran into this. I can take a look at if and create a PR for it if you like?

[refack]

I just ran into this. I can take a look at if and create a PR for it if you like?

I thought about doing it myself, but be my guest :)

[rvagg]

VersionSelectorScript.groovy:

  [ /sharedlibs_withoutssl/,          anyType,     lt(10)  ],

So that's allowing it to run on 10+. Should we be switching that to lt(12) since that change isn't backported?

[richardlau]

Having this job would allow us to catch things like nodejs/node#26182 (most likely due to a recent eslint update).

[refack]

Turned it on https://ci.nodejs.org/job/node-test-commit-linux-containered/nodes=ubuntu1604_sharedlibs_withoutssl_x64/10746/ but blocked by nodejs/node#26182

If someone could ping me when it's solved, I'll turn it back on.

[danbev]

@refack We have merged nodejs/node#26182 now. Would you be able to turn the job back on? Thanks!

[richardlau]

Might be worth holding off until the security releases are done just to be on the safe side: #1699

[richardlau]

Security releases are done. Ping @refack I think we can turn this on now.

[refack]

Turned on. Let's see https://ci.nodejs.org/job/node-test-commit-linux-containered/10929/

[refack]

https://ci.nodejs.org/job/node-test-commit-linux-containered/10929/nodes=ubuntu1604_sharedlibs_withoutssl_x64/ is ✔️ .
Closing.

[danbev]

Awesome, thanks @refack, @rvagg, and @richardlau!

[rvagg]

failing on v10.x with test-cli-node-print-help still https://ci.nodejs.org/view/All/job/node-test-commit-linux-containered/nodes=ubuntu1604_sharedlibs_withoutssl_x64/10933/

fix needs to be backported or we need to update jenkins/scripts/VersionSelectorScript.groovy to do lt(11) rather than lt(10)

[danbev]

I can take a look at backporting but it won't be until Monday at the earliest I'm afraid.

[rvagg]

@danbev ok, so let's leave the VersionSelectorScript with lt(10) for now anticipating getting it fixed early next week. But @BethGriggs will need to be aware of the failing test (test-cli-node-print-help on node-test-commit-linux-containered / withoutssl) that can be ignored while preparing for 10.15.3 in the meantime.

[danbev]

@rvagg It looks like the fix for this is already on v10.x-staging. I'm probably missing something obvious here (I'm a little sleep deprived and home with a sick kid today).

[richardlau]

@rvagg It looks like the fix for this is already on v10.x-staging. I'm probably missing something obvious here (I'm a little sleep deprived and home with a sick kid today).

The build Rod referenced was against v10.15.2. The fix was included in the v10.15.3 release that went out yesterday.

[richardlau]

There is, however, a failure on v11.x that we're not seeing on neither master nor v10.x: nodejs/node#26480