Date: Tue, 18 Sep 2007 21:26:26 -0400 From: "Bernie Volz \(volz\)" <volz@cisco.com> Subject: [dhcptest:00212] Re: DHCPv6 self-test RAND of .1 fails To: <dhcptest@tahi.org> Message-Id: <8E296595B6471A4689555D5D725EBB2105183752@xmb-rtp-20a.amer.cisco.com> In-Reply-To: <7ECEF9368E169544B43882BC98EB5D92032DAA08@SVR-ALH-EXC-02.mgc.mentorg.com> X-Mail-Count: 00212I'd say that the Tahi tests should allow EITHER. Part of having a RANDis just that - to make it random. If some client implementationinterprete it as inclusive and others exclusive, all the better (adds abit more randomness into the mix!). Personally, I would haveinterpretered as inclusive. - Bernie________________________________From: Leino, Tammy [mailto:tammy_leino@mentor.com] Sent: Tuesday, September 18, 2007 5:27 PMTo: dhcptest@tahi.orgSubject: [dhcptest:00211] Re: DHCPv6 self-test RAND of .1 failsI apologize for the multiple emails, but I wanted to mention that thetests do not fail when RAND is -.1, which leads me to believe there isan anomaly in the TAHI test suite when RAND is .1.Best Regards,Tammy Leino -----Original Message----- From: Leino, Tammy Sent: Tuesday, September 18, 2007 5:19 PM To: 'dhcptest@tahi.org' Subject: DHCPv6 self-test RAND of .1 fails Hello All, Whenever my node generates a RAND value of .1, the TAHI testsuite fails for all max retransmission time tests. It seems the TAHItest suite is interpretting the definition of RAND in RFC 3315 as -.1 to.1 non-inclusive. Here is the situation for RFC 3315 test #37 when RAND is chosenas .1 in retransmissions: First retransmission RT = IRT + (RAND * IRT) RT = 1 + (.1 * 1) RT = 1.1 Second retransmission RT = (2 * RTprev) + (RAND * RTprev) RT = (2 * 1.1) + (.1 * 1.1) RT = 2.31 Third retransmission RT = (2 * RTprev) + (RAND * RTprev) RT = (2 * 2.31) + (.1 * 2.31) RT = 4.851 The test fails here due to "Actual RT is bigger than theexpected maximum value". I am not sure what the expected maximum valueis. First, can someone comment on whether the TAHI test suiteinterprets the RAND values defined in RFC 3315 as inclusive ornon-inclusive? Second, if the values are supposed to be inclusive, can someoneverify that this is a problem with the TAHI test suite? Best Regards, Tammy Leino212_2.html (attatchment)(tag is disabled)