fix broken healthcheck tests

Four of the healthcheck tests were completely broken. They
were written with the option '--healthcheck-cmd' which is
not an option (it should be  '--healthcheck-command', with
'command' as a full word). The tests were merely checking
exit code, not error message, so of course they failed.
I have fixed the command line and added checks for the
expected diagnostic.

(Side note: do not write tests that check exit code but
nothing else. This should not need to be said).

One of the four tests was invalid: --healthcheck-interval 0.5s.
Per Brent:

   initially i was going to restrict sub one-second intervals

That test has been removed. It would probably be a good idea
for a future PR to add some validation such as preventing
negative values, but that's left as an exercise for later.

Also: grammar fix in an error message.

Caught by my ginkgo log greasemonkey script, which
highlights 'Error' messages and grabbed my attention.

Signed-off-by: Ed Santiago <santiago@redhat.com>

1 files changed, 1 insertions, 1 deletions

diff --git a/cmd/podman/shared/create.go b/cmd/podman/shared/create.go
index 7cf230605..eee5f515d 100644
--- a/cmd/podman/shared/create.go
+++ b/cmd/podman/shared/create.go
@@ -811,7 +811,7 @@ func makeHealthCheckFromCli(c *GenericCLIResults) (*manifest.Schema2HealthConfig
 		return nil, errors.Wrapf(err, "invalid healthcheck-start-period %s", inStartPeriod)
 	}
 	if startPeriodDuration < time.Duration(0) {
-		return nil, errors.New("healthcheck-start-period must be a 0 seconds or greater")
+		return nil, errors.New("healthcheck-start-period must be 0 seconds or greater")
 	}
 	hc.StartPeriod = startPeriodDuration