Handshake/WithHandshakeArgs4.RandomLoss/23 fail

[ami-GS]

Describe the bug

QUIC_CONNECTION_EVENT_SHUTDOWN_INITIATED_BY_TRANSPORT before checking Server->GetIsConnected() Simply slow machine? then how to manage this. Can we make idle timer longer for test? https://github.com/microsoft/msquic/actions/runs/9281377624/job/25537479115?pr=4327

Affected OS

• [X] Windows
• [ ] Linux
• [ ] macOS
• [ ] Other (specify below)

Additional OS information

No response

MsQuic version

main

Steps taken to reproduce bug

not repro locally

Expected behavior

pass

Actual outcome

D:\a\msquic\msquic\src\test\lib\HandshakeTest.cpp(323): error: Server->GetIsConnected() not true

Additional details

No response

[ami-GS]

OK, the timer is 10 sec which was set at 44. then expired at 54

06:50:44.810 [Microsoft-Quic][conn][0x243EF6A38B0] Setting TIMER.IDLE , delay=10000000 us

[ami-GS]

@nibanks

What's this? this says ignoring, but EventConnectionComplete is set which unblock the main thread then fail. https://github.com/microsoft/msquic/blob/cd5019f2de4251d45cda8783055165c7d32be27c/src/test/lib/TestConnection.cpp#L816C1-L822C89

[ami-GS]

My idea are

• make idle timeout longer
• retry connection if it there is idle timeout

[nibanks]

Random loss tests are hard. I don't know a way to guarantee they pass, because it's always theoretically possible (though rare) that all packets end up getting randomly dropped. Unless you have a good suggestion (I'll look at your PR), I think we just prioritize looking at other failed tests..... one idea, maybe we add retry support to our test scripts to retry test cases that fail?

[ami-GS]

how about retrying test from test.ps1 level?

[nibanks]

how about retrying test from test.ps1 level?

Sure, give it a try?

[nibanks]

I haven't seen this problem recently, especially since retry was added to the powershell.