-
Notifications
You must be signed in to change notification settings - Fork 10.2k
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
[release/2.1] Update branding to 2.1.28 #31547
Conversation
- version flows into `$(MicrosoftNETCoreApp21PackageVersion)` automatically - cherry-picked from #30729 internal branch
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Overall LGTM, not sure if other changes are required to make this official.
@@ -100,4 +100,8 @@ Later on, this will be checked using this condition: | |||
@aspnet/signalr-protocol-msgpack; | |||
</PackagesInPatch> | |||
</PropertyGroup> | |||
<PropertyGroup Condition=" '$(VersionPrefix)' == '2.1.28' "> |
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.
Above changes are for 2.1.26
and this is .28
is that intentional?
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.
Ah so this is the AspNetCore
package version https://github.com/dotnet/aspnetcore/pull/31547/files#diff-e4bed5b736f205989dd4fdb6d78acfe9126577983d325d378ce91794d74e63c8R5
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.
Yes, my changes in build/dependencies.props are just to use the packages that are already public in rolling and PR builds. We'll move to 2.1.27 when updating the baselines on Patch Tuesday.
Thanks @TanayParikh I'll get this in tomorrow, when the branches are slated to open. |
@wtgodbe @JunTaoLuo not sure if the commits I just added are needed for 3.1 or 5.0. Decided they made sense here because the Docker images we were using were very old. But, this bulks up the PR. Let me know if you'd prefer I did some parts in separate PRs. |
Holding for a actions/runner-images#3120 fix (or at least a workaround). Leaving this alone to make it easier for Azure / GitHub Actions team to see the problem. Will open another PR without Ubuntu upgrade in a bit… |
471a5ce
to
fde8f9a
Compare
/fyi this may take a bit more time because
|
fde8f9a
to
f0cf5f3
Compare
@@ -29,7 +30,8 @@ RUN apt-get update && \ | |||
lldb-3.6 \ | |||
wget && \ |
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.
If this long list can be pared down because the Ubuntu 18.04 image contains them already, I'll do that in #30729
@@ -96,6 +96,36 @@ public async Task FallbackSrcContent_Matches_CDNContent(ScriptTag scriptTag) | |||
Assert.Equal(RemoveLineEndings(cdnContent), RemoveLineEndings(fallbackSrcContent)); | |||
} | |||
|
|||
class RetryHandler : DelegatingHandler |
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.
Surprisingly, it's the tests in src/Identity/test/Identity.Test/CdnScriptTaghelperTests.cs that have been flaky in this PR. The RetryHelper
should be helpful in any case.
- skip `SharedFxTests.BaselineTest(...)` between rebranding and baseline updates - previous runtime is not available in this window - reduce retries in `RetryHelper` - download will usually fail repeatedly if it fails even once - previous maximum duration (105 min.) was greater than the job timeout - avoid NETSDK1023 warnings - eng/targets/CSharp.Common.props adds packages removed from project - create archives based on one that actually exists between rebranding and baseline updates - mostly cherry-picked from #30729 internal branch nit: remove unused `GetTotalRetriesUsed()` method
- set locale consistently on all platforms - default locale on newer agents is unloved `C.UTF-8`
- run `locale-gen` to avoid "setlocale: LC_ALL: cannot change locale" and similar - needs `locales` package - move to still-supported Ubuntu 18.04 image - image needs slightly-newer `libicu60`, `clang` and `lldb` packages - rename Dockerfile to match - cherry-picked from #30729 internal branch for Ubuntu image update nit: set locale-related environment variables to match Python default
- avoid MSB4011 warnings (about repeated Directory.Build.targets imports) - place `$DOTNET_HOME` beside repo root, not under it - cherry-picked from #30729 internal branch for image update nits: - use current Docker image for 2.1 alpine - set `$LANGUAGE` with other locale-related environment variables
- minor change shouldn't affect submodule build but best to keep up
- apply efe9b95 to similar template test
3d58e38
to
10e5c3b
Compare
/fyi I've been testing internal builds and addressing failures there. Test failures in PR validation builds appear to be flaky tests
|
Still blocked on
I'm working to duplicate @javiercn's 3368ea6 / #13646 fix but can't get https://github.com/dotnet/aspnetcore/blob/release/3.1/src/Identity/test/Identity.Test/RetryHandler.cs#L57 to compile in this branch. Suspect updating https://github.com/dotnet/aspnetcore/blob/release/2.1/eng/targets/CSharp.Common.props#L4 to |
/btw https://dev.azure.com/dnceng/internal/_build/results?buildId=1077719 confirmed our official builds (where we don't test 'natch) will succeed once this is in. |
@karelz @scalablecory - any ideas on the failure above in Details. It's coming from here, and only seems to be failing in 2.1. @dougbu I don't think you can update to compile with C# 8, because that requires a .NET SDK of 3.0 or later. What is the error you are seeing? |
In local runs with C# 7.2, I saw
To work around this, I ended up multiplying an |
Successful test results don't show much information but could at least see success of |
To be pendantic, that's not a C# compiler change, that's about the set of operators on |
Thanks for the additional context @Pilchie; I appreciate the pedantry. Also asked whether this might be specific to the BuildPool.Server.Amd64.VS2017.Open queue in https://teams.microsoft.com/l/message/19:[email protected]/1617924854323?tenantId=72f988bf-86f1-41af-91ab-2d7cd011db47&groupId=4d73664c-9f2f-450d-82a5-c2f02756606d&parentMessageId=1617924854323&teamName=.NET%20Core%20Eng%20Services%20Partners&channelName=First%20Responders&createdTime=1617924854323 |
@karelz @scalablecory @Pilchie Problem might be specific to .NET Framework runs and look relatively new. I saw it succeed on Monday in this branch (https://dev.azure.com/dnceng/public/_build/results?buildId=1073196) but it looked flaky in my internal builds for #30729 last week (https://dev.azure.com/dnceng/internal/_build?definitionId=21&branchFilter=150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570%2C150570). I'm going to try skipping on .NET Framework. But, would very much like to know why this test went South. |
- failures appear specific to that platform - leave src/Templating/test/Templates.Test/CdnScriptTagTests.cs unchanged - test project does not target .NET Framework
[Fact] | ||
// Because this test runs on .NET Core and seems reliable there, likely not worth running on .NET Framework. | ||
[ConditionalFact] | ||
[FrameworkSkipCondition(RuntimeFrameworks.CLR, SkipReason = "At least flaky on .NET Framework.")] |
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.
@Pilchie @dotnet/aspnet-build thoughts about filing a follow-up issue❔ I'm leaning toward no because the test is about external resources. Also realized the test runs fine on Linux and macOS.
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.
I think we should at least file an issue about figuring out why the test failed
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.
If that's in dotnet/runtime, so be it
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.
@wtgodbe I agree and wasn't really questioning that. My question was more whether we needed an issue about removing the [FrameworkSkipCondition(...)]
. Hearing none 😃
@karelz @scalablecory where should I file an issue about the .NET Framework HttpClient
class and the underlying SSL libraries❔
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.
@dougbu I believe we do not run dotnet/runtime tests on .NET Framework anymore. Before we stopped to run them, we saw quite some flakiness on .NET Framework in networking. I wonder if this is similar ...
If you're able to isolate it to HttpClient/SslStream repro, we can take a look -- I would start email thread with our team as a starting point.
CI is green now 🎉 Just waiting for a response on adding an issue tracking restoring the skipped .NET Framework test. (Would link to the issue in the new Just checked and MusicStore.csproj targets only .NET Core in 'release/3.1' and 'release/5.0', which might explain why we haven't seen failures in those branches. |
$(MicrosoftNETCoreAppPackageVersion)
$(MicrosoftNETCoreApp21PackageVersion)
automaticallySharedFxTests.BaselineTest(...)
between rebranding and baseline updatesRetryHelper
nit: remove unused
GetTotalRetriesUsed()
method