Minor problem with 64bTT: monthly accounting figures

[Garance A Drosihn]

At 3:57 PM -0700 4/19/04, Kris Kennaway wrote:
>On Mon, Apr 19, 2004, Garance A Drosihn wrote:
>  > At 2:08 PM -0700 4/19/04, Peter Wemm wrote:
>  > >
>>  >However, there is also a big scary comment that says:
>  > >
>  > >     * With sparc64 using 64-bit time_t's, there is some system
>  > >     * routine which sets ut_time==0 (the high-order word of a
>  > >     * 64-bit time) instead of a 32-bit time value.
>  > >
>>  >It sounds like something clobbers ut_time..
>>
>>  Big scary comment added by me, when fixing 'ac' to do more
>>  reasonable things with such records...  Afaik, we have still
>>  not figured out what it is that writes records with zero for
>>  the timestamp.
>
>Should an erratum be added in case this is unresolved by 5.3,
>or is this too minor an issue?

I have been considering it a minor issue.  But then, I also hoped
that someone would notice the new warning message from `ac', rerun
it in debug-mode, and be able to match up the bad-records with
whatever they were doing at the time.  And then we'd have a good
clue as to which program is sometimes writing these bad records.
But I guess that in normal operation, people won't see that message
until the monthly-run, so maybe it won't come up often.

In the cases I saw, it was just a few bad records over the course
of a month, so I assume that whatever-it-is, it is something that
doesn't happen often.  I haven't seen it happen on my systems at
all, but then almost all my sessions are made via ssh, so in my
case it would either happen all the time, or it would probably
never come up.

-- 
Garance Alistair Drosehn            =   gad at gilead.netel.rpi.edu
Senior Systems Programmer           or  gad at freebsd.org
Rensselaer Polytechnic Institute    or  drosih at rpi.edu