Some tests intermittently fail to terminate on OSX

[jen20]

When running ./run_tests.sh on Mac OS X, some tests (currently not sure which) intermittently do not terminate, instead enqueuing messages such as this ad infinitum:

[00828,15,23:14:40.447,DEBUG] EventStoreConnection 'ESC-109': enqueueing message EventStore.ClientAPI.Core.HandleTcpPackageMessage..
[00828,15,23:14:40.447,DEBUG] EventStoreConnection 'ESC-109': HandleTcpPackage connId 96996068-9a59-4527-8673-38adde53d497, package HeartbeatResponseCommand, 3389a251-8a60-4b6a-a163-c5f3f1e0e64a..

[gregoryyoung]

That is just heartbeat messages over the connection that is still alive.

On Tue, Aug 19, 2014 at 8:56 PM, James Nugent notifications@github.com
wrote:

When running ./run_tests.sh on Mac OS X, some tests (currently not sure
which) intermittently do not terminate, instead enqueuing messages such as
this ad infinitum:

[00828,15,23:14:40.447,DEBUG] EventStoreConnection 'ESC-109': enqueueing message EventStore.ClientAPI.Core.HandleTcpPackageMessage..
[00828,15,23:14:40.447,DEBUG] EventStoreConnection 'ESC-109': HandleTcpPackage connId 96996068-9a59-4527-8673-38adde53d497, package HeartbeatResponseCommand, 3389a251-8a60-4b6a-a163-c5f3f1e0e64a..

—
Reply to this email directly or view it on GitHub
#194.

Studying for the Turing test

[jen20]

Yes, the question is what is stopping it closing in the first place. I suspect it's a similar issue that we've seen before (on Linux) where the timeout on a close doesn't actually work.

[jen20]

This is highly intermittent. I can't figure out what is going on - sometimes the full suite will pass, sometimes a single test will fail. My guess is that a wait handle isn't being hit so MiniNodes aren't being torn down correctly.

[jen20]

Pushing this to v4.0.0. I have no idea wtf nunit is doing underneath on OS X.

[gregoryyoung]

Any update on this? Have you run nunit with trace so you can see which tests are the ones causing the issue?

[pgermishuys]

I haven't seen this on any of the officially supported platforms. Have seen flapping tests which are handled on a case by case basis, but since OSX isn't a supported platform, is this still relevant?