Flaky: Profiler_ThrowingOnSessionStartup_DoesntBreakSentryInit about sentry-dotnet HOT 6 CLOSED

I've looked into this multiple times and logically, the test seems sound to me - I have no clue how that condition could be failing. Can someone else have a look at the code? I'm likely missing some c#-related detail that comes into play here.

from sentry-dotnet.

Related:

• #3009
• #3002

from sentry-dotnet.

Still getting this all over the place (e.g. here). I get this pretty frequenly on my local machine as well (so not just a CI thing).

I think we should disable this test until we can work out what's going on and fix it.

from sentry-dotnet.

I've looked into this multiple times and logically, the test seems sound to me - I have no clue how that condition could be failing. Can someone else have a look at the code? I'm likely missing some c#-related detail that comes into play here.

Hm, as you say, logically it all looks OK.

The only theoretical situation in which I could imagine the test doing what it's doing is if this line doesn't throw an exception:

In that scenario, the constructor might return an instance before any exception from the Task.Wait was thrown... I guess that could happen if the work being done by _sessionTask didn't get scheduled within the timeout (in this case 10 seconds, so that seems highly improbably). And even if that was happening, we'd expect to see the log message from:

from sentry-dotnet.

I get this pretty frequenly on my local machine as well (so not just a CI thing).

One caveat in the comment on the PR... we're not expecting that PR to resolve this outside of CI. If this is resolved in CI but continues to be a problem on local machines, that at least confirms the source of the problem and we can circle back to remove the TestEnvironment check (factory.StartupTimedOut.Should().BeTrue() should be enough):

from sentry-dotnet.

We haven't had any trouble for a while, but did run into this today:

Nothing to do yet, I don't think... once could just be some twighlight zone anomaly but if it starts happening regularly again, we may need to reopen the case.

from sentry-dotnet.