remediate flaky timeout tests

Author: aldas

middleware/timeout_test.go

@@ -375,7 +375,7 @@ func TestTimeoutWithFullEchoStack(t *testing.T) {
 
 // NOTE: timeout middleware is first as it changes Response.Writer and causes data race for logger middleware if it is not first
 e.Use(TimeoutWithConfig(TimeoutConfig{
-Timeout: 15 * time.Millisecond,
+Timeout: 100 * time.Millisecond,
 }))
 e.Use(Logger())
 e.Use(Recover())
@@ -403,8 +403,13 @@ func TestTimeoutWithFullEchoStack(t *testing.T) {
 }
 if tc.whenForceHandlerTimeout {
 wg.Done()
+// extremely short periods are not reliable for tests when it comes to goroutines. We can not guarantee in which
+// order scheduler decides do execute: 1) request goroutine, 2) timeout timer goroutine.
+// most of the time we get result we expect but Mac OS seems to be quite flaky
+time.Sleep(50 * time.Millisecond)
+
 // shutdown waits for server to shutdown. this way we wait logger mw to be executed
-ctx, cancel := context.WithTimeout(context.Background(), 150*time.Millisecond)
+ctx, cancel := context.WithTimeout(context.Background(), 200*time.Millisecond)
 defer cancel()
 server.Shutdown(ctx)
 }