GC\LargeMemory\API\gc\gettotalmemory\gettotalmemory.cmd failed in rolling with "Large Memory Machine required"

[jakobbotsch]

Job: https://dev.azure.com/dnceng/public/_build/results?buildId=1760633&view=results
Test: https://dev.azure.com/dnceng/public/_build/results?buildId=1760633&view=ms.vss-test-web.build-test-results-tab&runId=47405440&resultId=101792&paneView=debug
Log: https://helixre8s23ayyeko0k025g8.blob.core.windows.net/dotnet-runtime-refs-heads-main-3e33c16d75004b5c98/PayloadGroup0/1/console.c2d85ac2.log?helixlogtype=result

The log shows:

  Starting:    GC.LargeMemory.XUnitWrapper (parallel test collections = on, max threads = 4)
    GC\LargeMemory\API\gc\gettotalmemory\gettotalmemory.cmd [FAIL]
      
      Return code:      1
      Raw output file:      C:\h\w\A70D08E8\w\B10209D6\uploads\Reports\GC.LargeMemory\API\gc\gettotalmemory\gettotalmemory.output.txt
      Raw output:
      BEGIN EXECUTION
       "C:\h\w\A70D08E8\p\corerun.exe" -p "System.Reflection.Metadata.MetadataUpdater.IsSupported=false"  gettotalmemory.dll 2048 
      Large Memory Machine required
      Test failed
      Expected: 100
      Actual: 0
      END EXECUTION - FAILED
      FAILED
      Test Harness Exitcode is : 1
      To run the test:
      > set CORE_ROOT=C:\h\w\A70D08E8\p
      > C:\h\w\A70D08E8\w\B10209D6\e\GC\LargeMemory\API\gc\gettotalmemory\gettotalmemory.cmd
      Expected: True
      Actual:   False
      Stack Trace:
           at GC_LargeMemory._API_gc_gettotalmemory_gettotalmemory_._API_gc_gettotalmemory_gettotalmemory_cmd()
      Output:
        
        Return code:      1
        Raw output file:      C:\h\w\A70D08E8\w\B10209D6\uploads\Reports\GC.LargeMemory\API\gc\gettotalmemory\gettotalmemory.output.txt
        Raw output:
        BEGIN EXECUTION
         "C:\h\w\A70D08E8\p\corerun.exe" -p "System.Reflection.Metadata.MetadataUpdater.IsSupported=false"  gettotalmemory.dll 2048 
        Large Memory Machine required
        Test failed
        Expected: 100
        Actual: 0
        END EXECUTION - FAILED
        FAILED
        Test Harness Exitcode is : 1
        To run the test:
        > set CORE_ROOT=C:\h\w\A70D08E8\p
        > C:\h\w\A70D08E8\w\B10209D6\e\GC\LargeMemory\API\gc\gettotalmemory\gettotalmemory.cmd

Note that we are running these in parallel and IIUC, they allocate a large chunk of memory. @dotnet/gc, is it ok to change these LargeMemory tests to succeed instead of fail on OutOfMemoryException? Or otherwise maybe we can force them to run sequentially instead?

[ghost]

Tagging subscribers to this area: @dotnet/gc
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Job: https://dev.azure.com/dnceng/public/_build/results?buildId=1760633&view=results
Test: https://dev.azure.com/dnceng/public/_build/results?buildId=1760633&view=ms.vss-test-web.build-test-results-tab&runId=47405440&resultId=101792&paneView=debug
Log: https://helixre8s23ayyeko0k025g8.blob.core.windows.net/dotnet-runtime-refs-heads-main-3e33c16d75004b5c98/PayloadGroup0/1/console.c2d85ac2.log?helixlogtype=result

The log shows:

  Starting:    GC.LargeMemory.XUnitWrapper (parallel test collections = on, max threads = 4)
    GC\LargeMemory\API\gc\gettotalmemory\gettotalmemory.cmd [FAIL]
      
      Return code:      1
      Raw output file:      C:\h\w\A70D08E8\w\B10209D6\uploads\Reports\GC.LargeMemory\API\gc\gettotalmemory\gettotalmemory.output.txt
      Raw output:
      BEGIN EXECUTION
       "C:\h\w\A70D08E8\p\corerun.exe" -p "System.Reflection.Metadata.MetadataUpdater.IsSupported=false"  gettotalmemory.dll 2048 
      Large Memory Machine required
      Test failed
      Expected: 100
      Actual: 0
      END EXECUTION - FAILED
      FAILED
      Test Harness Exitcode is : 1
      To run the test:
      > set CORE_ROOT=C:\h\w\A70D08E8\p
      > C:\h\w\A70D08E8\w\B10209D6\e\GC\LargeMemory\API\gc\gettotalmemory\gettotalmemory.cmd
      Expected: True
      Actual:   False
      Stack Trace:
           at GC_LargeMemory._API_gc_gettotalmemory_gettotalmemory_._API_gc_gettotalmemory_gettotalmemory_cmd()
      Output:
        
        Return code:      1
        Raw output file:      C:\h\w\A70D08E8\w\B10209D6\uploads\Reports\GC.LargeMemory\API\gc\gettotalmemory\gettotalmemory.output.txt
        Raw output:
        BEGIN EXECUTION
         "C:\h\w\A70D08E8\p\corerun.exe" -p "System.Reflection.Metadata.MetadataUpdater.IsSupported=false"  gettotalmemory.dll 2048 
        Large Memory Machine required
        Test failed
        Expected: 100
        Actual: 0
        END EXECUTION - FAILED
        FAILED
        Test Harness Exitcode is : 1
        To run the test:
        > set CORE_ROOT=C:\h\w\A70D08E8\p
        > C:\h\w\A70D08E8\w\B10209D6\e\GC\LargeMemory\API\gc\gettotalmemory\gettotalmemory.cmd

Note that we are running these in parallel and IIUC, they allocate a large chunk of memory. @dotnet/gc, is it ok to change these LargeMemory tests to succeed instead of fail on OutOfMemoryException? Or otherwise maybe we can force them to run sequentially instead?

Author:	jakobbotsch
Assignees:	-
Labels:	area-GC-coreclr, untriaged
Milestone:	-

[mangod9]

yeah would make sense to make these tests more reliable by possibly ignoring OOMs. Not sure if there is a way to mark these as "skipped" within AzDO?

[AntonLapounov]

Hit this in my PR: https://dev.azure.com/dnceng/public/_build/results?buildId=1904603&view=ms.vss-test-web.build-test-results-tab.