[1.0.44]
When I watch my Processing Queue window, NCrunch seems to queue a lot of "sessions" with only 10 tests, where expectation of each session run is about ~12 seconds.
Why does not NCrunch queue more tests in each "session", and can I somehow tweak it to push more tests in one session? If not, is such feature scheduled?
(Some tests require factory initializations which takes about 10s, using [ExclusiveUses] attribute does not seem to "help" either. Having ~5000 tests and NCrunch pools 10tests á 10sec off course makes the total test run to take forever...).
Thanks.
Remco NCrunch Developer
#3726
19 Feb 2013 22:10 UTC
Hi,
The test pipeline builder takes a huge range of different factors into consideration, including the running time of individual tests, the running time of their fixtures, their impact status, failure status, resource usage, etc.
I'm aware of a problem/defect at the moment in the pipeline where sometimes it can queue up tests that are fast running but with a collectively slow fixture setup inside separate tasks if these fast running tests have notable differences in their priority (for example, if 2/10 of them have failed, they may be run separately). This isn't intentional and I'm hoping to have it sorted out soon.
When you see the tasks showing tests that you think should be run together, what is the priority difference between these tasks in the processing queue?
Also - What is the mechanism you are using for your factory initializations? Is this done via test fixture setup?
The test pipeline builder takes a huge range of different factors into consideration, including the running time of individual tests, the running time of their fixtures, their impact status, failure status, resource usage, etc.
I'm aware of a problem/defect at the moment in the pipeline where sometimes it can queue up tests that are fast running but with a collectively slow fixture setup inside separate tasks if these fast running tests have notable differences in their priority (for example, if 2/10 of them have failed, they may be run separately). This isn't intentional and I'm hoping to have it sorted out soon.
When you see the tasks showing tests that you think should be run together, what is the priority difference between these tasks in the processing queue?
Also - What is the mechanism you are using for your factory initializations? Is this done via test fixture setup?
Post a reply
Log in to reply.