Uname borked on ??-Release...

[Kevin Kinsey]

Kris Kennaway wrote:
> Kevin Kinsey wrote:
> 

<snip>

>> I get the following from uname -a:
>>
>> FreeBSD archangel.daleco.biz 6.2-RELEASE FreeBSD 6.2-RELEASE #6:
>> Sat Jun  2 09:22:50 CDT 2007  root at archangel.daleco.biz:
>> /usr/obj/backup/src/sys/GENERIC  i386
>>
>> However, I rebuilt world, more or less without issues,
>> twice in February with "RELENG_6" in the supfile.  This
>> didn't change uname's output, and that worried me a bit.
>>
>> So, to make matters bette^H^H^H^Hadder, I csup'ped
>> to RELENG_7_0 the day after it was released, read
>> /usr/src/UPDATING, and the webpage detailing the
>> upgrade, and did another buildworld/kernel cycle.
>> Now I have no idea if I'm on 6 or 7 (seems like
>> 7, but many ports issues, and I've rebuilt them
>> all), and it's just becoming a major PITA.
> 
> 
> You didnt succeed in installing the new kernel.  'make installkernel' is 
> the step in which this occurs.
> 
> Kris

Thank you and Phillip for answering my post.  However,
I've done this 3 times now, and I don't skip that step.
There have been no errors in the process, either.

AAMOF, in response to Phillip's mail, I just did it
again, as you can see (z* is to omit snipping):

ll /boot/kernel/z*
-r-xr-xr-x  1 root  wheel   712006 Mar  3 15:16 /boot/kernel/zfs.ko*
-r-xr-xr-x  1 root  wheel  3471592 Mar  3 15:16 /boot/kernel/zfs.ko.symbols*
-r-xr-xr-x  1 root  wheel    38175 Mar  3 15:16 /boot/kernel/zlib.ko*
-r-xr-xr-x  1 root  wheel    58834 Mar  3 15:16 /boot/kernel/zlib.ko.symbols*

I've rebooted the system, and I'm still being told I'm running
6.2 by uname.  In addition, pkg_add thinks I should be looking
for 6-latest packages instead of 7, and the list of annoyances
continues.  And, "hmm", symbols?  I'm guessing that knob is ON
in FBSD7?  Once again, proof that something's wrong, as I didn't
build debugging kernels in FBSD6 ... so I'm thinking this is
a 7 kernel?  It just doesn't make sense to me.

It *is* a Monday, after all.  

If installkernel didn't succeed, shouldn't there be any
other evidence?  Could skipping a mergemaster at some point
have this effect?  (I don't *always* do that, unless I'm 
making a pretty big move, and the first build cycle was
production code IIRC)

What about issues with "newvers.sh" (or whatever it is?)
Any other think-outside-the-box stuff?  What could cause
an installkernel operation to fail but appear to succeed?

KDK
-- 
I just rewrote my .sig.