TestResolvDNS#test_no_server always fails on usa's mswin environment
Every week usa (Usaku NAKAMURA) reports a test failure on his mswin environment like this:
[ 9/135] TestResolvDNS#test_no_server = 5.00 s 1) Error: TestResolvDNS#test_no_server: Timeout::Error: execution expired C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:672:in `wait_readable' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:672:in `request' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:526:in `block in fetch_resource' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:1120:in `block (3 levels) in resolv' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:1118:in `each' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:1118:in `block (2 levels) in resolv' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:1117:in `each' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:1117:in `block in resolv' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:1115:in `each' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:1115:in `resolv' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:517:in `fetch_resource' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:507:in `each_resource' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:498:in `getresources' C:/Users/usa/develop/ruby/core/mytree/test/resolv/test_dns.rb:167:in `block (2 levels) in test_no_server' C:/Users/usa/develop/ruby/core/mytree/lib/resolv.rb:300:in `open' C:/Users/usa/develop/ruby/core/mytree/test/resolv/test_dns.rb:166:in `block in test_no_server' C:/Users/usa/develop/ruby/core/mytree/lib/timeout.rb:108:in `timeout' C:/Users/usa/develop/ruby/core/mytree/test/resolv/test_dns.rb:165:in `test_no_server' Finished tests in 854.592396s, 23.4942 tests/s, 6046.6370 assertions/s. 20078 tests, 5167410 assertions, 0 failures, 1 errors, 328 skips ruby -v: ruby 2.7.0dev (2019-01-28 trunk 66933) [x64-mswin64_140] NMAKE : fatal error U1077: '.\ruby.exe' : リターン コード '0x1' Stop.
It started roughly after timer thread was eliminated in Ruby 2.6.
Note that this is not reproducible on mswin RubyCI.
normalperson (Eric Wong) Do you have any insight for a possible cause of this?
Updated by normalperson (Eric Wong) 14 days ago
FYI, I've never seen GREEN result of the test since last September.
The environment is Windows10 + Visual C++ 14.0(=Visual Studio 2015).
Do you have info on which revision it started failing with?
Timer-thread removal happened in August; and should only affect
thread_pthread; and mswin uses thread_win32, right?
Btw, feel free to email me directly for faster response.
I don't have time to follow Ruby closely, anymore.