Browse code

Extend sleep/timeout for ngoroutine check

In cases where this is failing it's ok to have the test take extra time,
but we don't want it to fail because of some race/performance
differences between systems.
So make the timeout go to 30s and double the sleep time in between
checks so it's not pounding the daemon quite so fast.

Originally I couldn't make this test fail (pre-change), but changed
graphdrivers and saw a failure pretty quickly.
This change seems to smooth that out.

Signed-off-by: Brian Goff <cpuguy83@gmail.com>

Brian Goff authored on 2015/05/15 22:54:43
Showing 1 changed files
... ...
@@ -432,7 +432,7 @@ func (s *DockerSuite) TestLogsFollowGoroutinesWithStdout(c *check.C) {
432 432
 	c.Assert(cmd.Process.Kill(), check.IsNil)
433 433
 
434 434
 	// NGoroutines is not updated right away, so we need to wait before failing
435
-	t := time.After(5 * time.Second)
435
+	t := time.After(30 * time.Second)
436 436
 	for {
437 437
 		select {
438 438
 		case <-t:
... ...
@@ -441,7 +441,7 @@ func (s *DockerSuite) TestLogsFollowGoroutinesWithStdout(c *check.C) {
441 441
 			if nroutines == getNGoroutines() {
442 442
 				return
443 443
 			}
444
-			time.Sleep(100 * time.Millisecond)
444
+			time.Sleep(200 * time.Millisecond)
445 445
 		}
446 446
 	}
447 447
 }