Update helix references#61755
Conversation
|
There are some new Kestrel test failures on Ubuntu22 here that we didn't previously see on Ubuntu20 - they look to be Crypto related. @halter73 can you route? |
|
Ubuntu 22.04 has OpenSSL 3.x. Perhaps the tests are not expecting that? These type of upgrades frequently cause breaks in dotnet/runtime. We're testing with Ubuntu 25.04 in that repo, with a goal of moving to 26.04 before it ships. Let's get to Ubuntu 22.04 first, but we should try to adopt the same model here as with runtime (in main). |
|
Discussing this offline w/ Stephen & @JeremyKuhne |
|
I pinged @bartonjs about it. He asked if we're using a certificate with a chain involving a UPN Name Constraint, but I don't think we are. The simplest one might be this one that just tries to new up an X509Certificate2 from a test aspnetdevcert.pfx that we pass directly to UseHttps() which AFICT doesn't have any unusual constraints. It's very expired, but that's a different issue.
aspnetcore/src/Servers/Kestrel/Kestrel/test/HttpsConfigurationTests.cs Lines 211 to 236 in 83e90c4 |
|
FWIW, if I create a "Ubuntu Server 22.04 LTS - x64 Gen2" VM on Azure, clone the aspnetcore repo and run the UseHttpsMayNotImplyUseKestrelHttpsConfiguration test, it passes. I think it's something going on with the Helix agent in particular and not just Ubuntu 22.04 in general. Should we try recreating the agent? Here's the failure on the CI: https://dev.azure.com/dnceng-public/public/_build/results?buildId=1031039&view=ms.vss-test-web.build-test-results-tab&runId=27684240&resultId=113534&paneView=debug |
|
We could also try running in a container. That might be better anyway. |
| <!-- aspnetcore-quarantined-pr (quarantined-pr.yml) --> | ||
| <ItemGroup> | ||
| <HelixAvailableTargetQueue Include="Ubuntu.2004.Amd64.Open" Platform="Linux" /> | ||
| <HelixAvailableTargetQueue Include="Ubuntu.2204.Amd64.Open" Platform="Linux" /> |
There was a problem hiding this comment.
I see you reverted the Ubuntu2004 -> 2204 change in Helix.targets - did you mean to revert this one as well?
There was a problem hiding this comment.
Yes. That one will require some investigation and change. I was wanting to see if we could make a clean PR w/o that change.
Co-authored-by: William Godbe <wigodbe@microsoft.com>
|
This looks clean. I see the E2E test failure in another PR. The helix run you asked for is clean: https://dev.azure.com/dnceng-public/public/_build/results?buildId=1037803&view=results Good to merge @wtgodbe? |
|
Yep! All failures are unrelated. Merging |
I tried to modernize how we handle helix references (and update them). These changes may require a couple of updates.
Related: