> How many sets of tests do you normally run in parallel?
What defines "sets of tests" ?
> there a particular point (in max execution threads) at which you really start to notice the engine falling over?
Yes, 2. As shown below though it isn't apparent when the issue is starting to appear. To be able to get some work done without waiting for NCrunch to become responsive I've disabled parallel testing during the work day for now.
A) Engine idle, force execution of 2 tests with "run in existing process" with settings: Fast Lane Threads(1), Max number of processing threads(2), Max number of test runner processes to pool(2), Allow parallel test execution(True), Allow tests to run in parallel with themselves
Code:
[07:16:34.5653-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3815381 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:16:39.9468-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3805380 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:16:45.4144-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.4645464 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:16:50.8129-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3265326 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:16:56.0394-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.2265226 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:17:01.575-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3935393 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:17:06.8895-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3135313 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
B) Engine idle, force execution of same 2 tests with "run in existing process" with same settings except Allow parallel test execution(False) (unfortunately I did not do this before step A to verify if the issue was there before switching to parallel or not).
Code:
[07:20:22.3491-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.4715471 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:20:27.6636-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3135313 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:20:32.9691-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3045304 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:20:38.4547-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.3985398 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
[07:20:43.6892-Core-94] Event [TestDataUpdatedEvent] took 00:00:05.2325232 to process on CORE thread when calling nCrunch.Client.Model.CoverageWatcher.testDataUpdated
Restarted vstudio (same no of documents open) and redid B)
<no output at all> (yes "log to output window" was set to true, got a lot of nullreferencexception in output window when closing solution which I submitted)
Restarted vstudio (same no of documents open) and redid A)
<no console output at all>
I will try to redo A) some time later today and verify if B) still outputs the long event durations.