Comment 3 for bug 1667055

Another thought. Could it be that it works as designed, and api users may expect the number of agents scheduled for an HA router changing in short time since its creation?

I see some other test cases, like test_keepalived_multiple_sighups_does_not_forfeit_mastership, do:

        common_utils.wait_until_true(
            lambda:
            len(self.client.list_l3_agent_hosting_routers(
                router['id'])['agents']) == 2,
            timeout=90)

instead. Should we do the same in test_ha_router?