Peter Firmstone created RIVER-469:
-------------------------------------

             Summary: Integration test 
org/apache/river/test/impl/lookupdiscovery/MulticastRequestDelay.td
                 Key: RIVER-469
                 URL: https://issues.apache.org/jira/browse/RIVER-469
             Project: River
          Issue Type: Test
          Components: net_jini_discovery
    Affects Versions: River_2.2.3, River_3.0.0, River_2.2.2, River_2.2.1, 
River_2.2.0, River_2.1.2, River_2.1.1
            Reporter: Peter Firmstone


The integration test 
org/apache/river/test/impl/lookupdiscovery/MulticastRequestDelay.td was never 
completed and is configured to be skipped by the test suite.

The intent of the test is to check the configuration of 
initialMulticastRequestDelayRange results in an average delay where the mean 
delay is halfway between zero and the configured value.

The test records the start time, then starts 100 LookupDiscovery instances, 
there is a significant start up penalty and a significant amount of time 
elapses by the time the last LookukpDiscovery instance has started.   
Additionally each LookupDiscovery continues to send multicast datagram packets 
and the test only measures the first 100 datagram packets that arrive from 
localhost.

The problem is that the test does not determine whether the packet received is 
the first packet sent by a LookupDiscovery instance.

Also the test should only check whether each received datagram is sent from a 
local network interface, as the packets aren't sent from localhost in this 
instance.

Recommend rewriting this test to start one LookupDiscovery at a time, measure 
the delay time to receive the first multicast datagram from that 
LookupDiscovery, then shut it down.

Also recommend that the LookupDiscovery delay be modified to account for it's 
own startup delay, so that the average result is as expected.

 
||•|| {{initialMulticastRequestDelayRange}}||
| ||Type:|{{long}}|
| ||Default:|{{0 milliseconds}}|
| ||Description:|With respect to when this utility is started, this entry 
controls how long to wait before sending out the first multicast request. If 
the value is positive, the first request will be delayed by a random value 
between {{0}} and {{initialMulticastRequestDelayRange}} milliseconds. 
Subsequent request intervals are controlled by the [ 
{{multicastRequestInterval}}|http://river.apache.org/release-doc/current/api/net/jini/discovery/LookupDiscovery.html#multicastRequestInterval]
 entry. Note that this entry only has effect when this utility is initialized. 
The first multicast request is not delayed if the groups to discover are 
subsequently changed.|



--
This message was sent by Atlassian Jira
(v8.3.2#803003)

Reply via email to