Date: Tue, 2 Sep 2008 16:14:52 -0700
From: Anthony Coon <acoon@vmware.com>
Subject: [users:00875] Re: failure3 .2.5.B: Part B: Prefix Lifetime greater than 2 hours(RFC 4862)
To: "users@tahi.org" <users@tahi.org>
Message-Id: <4C746D982532424D8D9F286C8A9B11E11B91199B8C@PA-EXMBX04.vmware.com>
In-Reply-To: <48BCA7DA.6030402@redhat.com>
References: <48A94161.2010302@redhat.com> <A8D19528FBF27C4691BD90C81EB22ABE04EBD9AB@rdlnexch03.marvell.com> <48B35492.1080800@redhat.com> <A8D19528FBF27C4691BD90C81EB22ABE04F551DA@rdlnexch03.marvell.com>,<48BB534C.6040206@redhat.com> <4C746D982532424D8D9F286C8A9B11E11B91199B87@PA-EXMBX04.vmware.com>,<48BBD6A9.4000903@redhat.com> <4C746D982532424D8D9F286C8A9B11E11B91199B8A@PA-EXMBX04.vmware.com>,<48BCA7DA.6030402@redhat.com>
X-Mail-Count: 00875

Wang,

That's a lot of drift.

I did a TAHI Core test of RHEL5.2 as NUT and freebsd 7.0 as TN and the test you noted passed.  Both RHEL5.2 and freebsd were running in virtual machines on the same server.  drift.pl 3600 on the freebsd VM reported 3600.

My guess is that your over sleep is the first thing you need to fix.  If my results are representative, the IPv6 stack is ok.

I patched vSleep to accomodate the clock drift on my similar machine since I don't have the time right now to trace down the problem in freebsd.

Good luck.

cheers,
Tony
________________________________________
From: wang_jiabo [jiabwang@redhat.com]
Sent: Monday, September 01, 2008 7:41 PM
To: users@tahi.org
Subject: [users:00874] Re: failure3 .2.5.B: Part B: Prefix Lifetime greater than 2 hours(RFC 4862)

Hello, Tony:
   Thanks, my TN computer get 3745, I think my TN should be correct the clock drift.
my NUT is RHEL5.2, and TN freebsd 6.2
Thanks
Wang jiabo

Anthony Coon wrote:

Wang,

It could be your clock or a defect in your ipv6 stack.

I use the following utility to measure sleep drift for tahi machines.  I have a HP box with 2 Xeons that drifts 10 seconds in 5 minutes with speedstep off.

If you are not seeing a material drift, you should probably take this up with your developers as the stack may not be handling RA addresses correctly.

===============
#!/usr/bin/perl

$tm=$ARGV[0];
print "tm=$tm\n";

$tm0=time();

sleep($tm);

$tm1=time();
$tmx=$tm1-$tm0;
print "clock delta: $tmx\n";
=================

Run this for an hour,  for example:
./drift.pl 3600

If it returns a delta materially different than 3600, you probably have a clock problem, else you may have a software defect in your stack... or both ;-)

BTW, is the NUT RHEL5.2 and TN freebsd 7.0?

HIH,
Tony

________________________________________
From: wang_jiabo [jiabwang@redhat.com<mailto:jiabwang@redhat.com>]
Sent: Monday, September 01, 2008 4:48 AM
To: users@tahi.org<mailto:users@tahi.org>
Subject: [users:00872] Re: failure3 .2.5.B: Part B: Prefix Lifetime greater than 2 hours(RFC 4862)

Hello, all:
very strange, I close the speedstep in Freebsd computer, but it is still
failure,
could you support me finding where problem is?

Anthony Coon wrote:


Wang,

Any chance that you are running on Intel chips with speedstep?  See

http://lists.freebsd.org/pipermail/freebsd-bugs/2007-February/022317.html

and

http://www.mavetju.org/mail/view_message.php?list=freebsd-bugs&id=2529282

HIH,
Tony



<snip>