|Subject:||'second request honored asynchronously' t/10_resolver.t test fails|
The 'second request honored asynchronously' t/10_resolver.t test fails on a slow host: $ prove -l -v t/10_resolver.t t/10_resolver.t .. 1..6 ok 1 - first request honored asynchronously ok 2 - first request not from cache ok 3 - first request from internal resolver not ok 4 - second request honored asynchronously ok 5 - second request not from cache ok 6 - second request from external resolver # Failed test 'second request honored asynchronously' # at t/10_resolver.t line 83. # Looks like you failed 1 test of 6. Dubious, test returned 1 (wstat 256, 0x100) Failed 1/6 subtests This is a race between reusing an TCP connection and a keep-a-live timeout set to 120 s in the library. The bug can be reproduced by adding "sleep 130;" into t/10_resolver.t at beginning of got_conn() subroutine. The 0.272 release reworded the message (second request connect error 110: Connection timed out), but the issue is still there. I recommend to skip the affected tests if the test run time exceeds the time-out.