Clustering tests failing with explicitly specified IPv6 addresses

Many HTTP session-based clustering tests are failing when the testsuite is started up with IPv6 addresses, for example:

./build.sh -Dnode0=[3ffe:ffff:0100:f101::1] -Dnode1=[3ffe:ffff:0100:f101::2] -DudpGroup=[ff0e::1:2:3] tests

When the testsuite is started using domain names which resolve to the same IPv6 addresses, the tests pass:

./build.sh -Dnode0=lenovo6a -Dnode1=lenovo6b -DudpGroup=[ff0e::1:2:3] tests

An example of a typical failure, in the test SimpleTestCase running under UDP:

Error Message
getSessionCookie(): fail to find session id. Server name: [fec0:0:a16:ffff::12]
Stacktrace
junit.framework.AssertionFailedError: getSessionCookie(): fail to find session id. Server name: [fec0:0:a16:ffff::12]
at org.jboss.test.cluster.test.BaseTest.getSessionCookie(BaseTest.java:269)
at org.jboss.test.cluster.test.BaseTest.setCookieDomainToThisServer(BaseTest.java:242)
at org.jboss.test.cluster.test.SimpleTestCase.testHttpSessionReplication(SimpleTestCase.java:81)
at junit.extensions.TestDecorator.basicRun(TestDecorator.java:24)
at junit.extensions.TestSetup$1.protect(TestSetup.java:21)
at junit.extensions.TestSetup.run(TestSetup.java:25)

The problem seems to be related to the use of cookies in the HTTP session replication tests.
Here is a message which is emitted during many of these session based tests:

2009-10-21 16:30:56,002 WARN [org.apache.commons.httpclient.HttpMethodBase] Cookie rejected: "$Version=0; JSESSIONID=AQYACiGZiyqiFpeNHDA1Ug**; $Path=/". Illegal domain attribute "[fec0". Domain of origin: "[fec0:0:a16:ffff::11]"

HttpClient seems to be having two problems with cookies:
it cannot correctly parse cookie domains which are IPv6 addresses (see https://issues.apache.org/jira/browse/HTTPCLIENT-654)
(ii) even if I fix the issue HTTPCLIENT-654 by modifying the codebase as described, it seems that cookies are still getting rejected - don't know if this is due to the cookie management policy

The cookie angle may be a red herring, but it was the only difference I could see between a clean run and a failing run.