Description
openedon May 22, 2023
I'm having a similar issue, and the build fails with the following trace running in Azure Devops hosted agents:
##[error]DiscoveryMessage : System.InvalidOperationException: The provided manager was not found in any slot.
at Microsoft.VisualStudio.TestPlatform.CrossPlatEngine.Client.ParallelOperationManager`3.ClearCompletedSlot(TManager completedManager)
at Microsoft.VisualStudio.TestPlatform.CrossPlatEngine.Client.ParallelOperationManager`3.RunNextWork(TManager completedManager)
at Microsoft.VisualStudio.TestPlatform.CrossPlatEngine.Client.Parallel.ParallelProxyDiscoveryManager.HandlePartialDiscoveryComplete(IProxyDiscoveryManager proxyDiscoveryManager, Int64 totalTests, IEnumerable`1 lastChunk, Boolean isAborted)
at Microsoft.VisualStudio.TestPlatform.CrossPlatEngine.Client.Parallel.ParallelDiscoveryEventsHandler.HandleDiscoveryComplete(DiscoveryCompleteEventArgs discoveryCompleteEventArgs, IEnumerable`1 lastChunk)
at Microsoft.VisualStudio.TestPlatform.CrossPlatEngine.Client.ProxyDiscoveryManager.InitializeDiscovery(DiscoveryCriteria discoveryCriteria, ITestDiscoveryEventsHandler2 eventHandler, Boolean skipDefaultAdapters)
at Microsoft.VisualStudio.TestPlatform.CrossPlatEngine.Client.ProxyDiscoveryManager.DiscoverTests(DiscoveryCriteria discoveryCriteria, ITestDiscoveryEventsHandler2 eventHandler)
The issue started when moving from 17.5 to 17.6 two days ago.
Originally posted by @jeromelaban in #4467 (comment)
Update: To clarify this issue for future implementation of a fix. When testhost crashes in initialization, we don't communicate back that initialization did not happen and try to continue to the next step, which again does initialization and fails again. This will trigger cleanup 2 times via CompletedRun event, which cleans slots and pushes next work to be executed. First the cleanup will succeed, but the second one will fail. This also highlights that there is a disconnect between the work being done, and when we think the work already completed.
