Ceph tests sometimes fail to contact rados gw

The ceph tests are blocked a bit by another utah bug right now, but it seems like sometimes they actually manage to run. I saw
http://10.98.0.1:8080/view/Raring/view/Smoke%20Testing/job/raring-server-amd64-smoke-ceph/33/artifact/clientlogs/utah-1928-raring-server-amd64_ceph.run_2012-12-14_08-12-17.yaml/*view*/ on a recent run with the following failures:

  ...FERROR:root:Failed to open localhost, 'HTTP Error 500: Internal Server Error' error
    F.
    ======================================================================
    FAIL: testCephRadosGateway (__main__.CephDaemonTest)
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "test.py", line 50, in testCephRadosGateway
        "Ceph Rados Gateway Daemon not running or to many daemons running: %s" % output)
    AssertionError: Ceph Rados Gateway Daemon not running or to many daemons running:

    ======================================================================
    FAIL: testCephRadosGatewayApache (__main__.CephDaemonTest)
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "test.py", line 58, in testCephRadosGatewayApache
        self.fail("Failed to connect to the localhost")
    AssertionError: Failed to connect to the localhost

    ----------------------------------------------------------------------

I'm not able to reproduce these locally when running it by hand, but I think the problem might be with the 20 second sleep at the end of setup.sh. I suspect that the reason I'm not seeing it consistently may be because I'm taking longer, giving the daemon a bit more time to start up. Could we try increasing this timeout, or setting the test to retry to see if that fixes these up?