Observed
TC build Test .NET Framework (X64, VS 17.0, .NET SDK 8.0) #19 (branch release/2026.0): the whole run is red because of a single build-test failure:
PostSharp.Patterns.Caching.BuildTests: CacheAttributeTests.CacheRefOut -- Test build failed.
Root cause (from the inner binlog msbuild-335b3d2e.binlog, published in failure-logs)
The test build's NuGet restore tried to download a package from nuget.org and the network failed:
error: Failed to download package 'System.Diagnostics.DiagnosticSource.4.6.0'
from 'https://api.nuget.org/v3-flatcontainer/system.diagnostics.diagnosticsource/4.6.0/...'
error: An error occurred while sending the request.
error: Unable to connect to the remote server.
error: A connection attempt failed because the connected party did not properly respond
after a period of time, or established connection failed because connected host has failed to respond
The test burned ~9 minutes in connection timeouts before failing. Nothing is wrong with the test or the weaver — this is a transient network outage on the agent surfacing as a test failure.
Why this deserves fixing rather than re-running
Build tests restore NuGet packages on the test path: any cache miss goes out to nuget.org, so a network blip anywhere in a ~5,000-test run turns the build red and costs a full re-run. The package set referenced by tests is pinned and known in advance.
Remediation options
- Pre-populate the package cache for all test-referenced packages during
Prepare/PrepareTestUtilities (or bake them into the agent/Docker image) so test-time restores never touch the network.
- Run test-time restores with the local cache as the only source (fail fast with a clear "infrastructure" message instead of 9 minutes of timeouts).
- As a stopgap, retry restore-class failures once in the testing framework before declaring the test failed.
🤖 Generated with Claude Code
Observed
TC build
Test .NET Framework (X64, VS 17.0, .NET SDK 8.0)#19 (branchrelease/2026.0): the whole run is red because of a single build-test failure:Root cause (from the inner binlog
msbuild-335b3d2e.binlog, published infailure-logs)The test build's NuGet restore tried to download a package from nuget.org and the network failed:
The test burned ~9 minutes in connection timeouts before failing. Nothing is wrong with the test or the weaver — this is a transient network outage on the agent surfacing as a test failure.
Why this deserves fixing rather than re-running
Build tests restore NuGet packages on the test path: any cache miss goes out to nuget.org, so a network blip anywhere in a ~5,000-test run turns the build red and costs a full re-run. The package set referenced by tests is pinned and known in advance.
Remediation options
Prepare/PrepareTestUtilities(or bake them into the agent/Docker image) so test-time restores never touch the network.🤖 Generated with Claude Code