Here’s one of my builds that took about 20 minutes to run. The main reason (that I’m concerned with) is that running NUnit tests pauses between test executions.
Locally, if I run dotnet test, it executes the tests in a few seconds (less than ten). If I run it with normal verbosity (like I do in Travis), I can see the individual test execution times are slightly slower than when they run on Travis; the total runtimes reported also seem similar.
However, I noticed that the build seems to freeze for several seconds between each successive test case. I suspected output buffering, but the builds actually do take very long.
Is this actually a problem and/or something I can do something about?
1 Like
https://travis-ci.org/rambotech/BOG.SwissArmyKnife/builds/651017725
I am having the same issue: running on my desktop, the test process is only taking 24 seconds. I don’t see any command line switches which would affect this. In fact, it doesn’t change if I add the --no-build switch to dotnet test command. Not sure why the wide variation in execution time is occurring.
I have the same problem on https://travis-ci.org/microting/eform-sdk-dotnet/builds/660086733
Ths takes 30 minutes in total to run on a local computer, but on Travis it it’s Total time 1 hr 58 min 42 sec
It’s probably related to the issue with dotnet test writes “Test run in progress…”
Profiling is the only way that I know of to find out where it’s spending time. Output buffering is possible – in this case, time would be spent blocked in console I/O. But since e.g. https://travis-ci.org/rambotech/BOG.SwissArmyKnife/builds/651017725 doesn’t generate much output (look at the raw log to also see output that may be hidden with terminal control characters), that’s improbable.