At 8:55 AM -0600 3/1/04, Tillman Hodgson wrote:
>Look a little odd this month:
>
>Subject: caliban.rospa.ca monthly run output
>
>Doing login accounting:
> root 0.84
> total -298848.27
> toor -298849.12
>
>-- End of monthly output --
>
>Luckily this isn't a box where I care much about accounting, but
>a Heads-Up in the conversion notes might be warranted.
Having checked into this, I am pretty sure that the real bug is
not with 'ac', but with code somewhere else in the system which
generates logoff records with ut_time==0. Still, 'ac' needs to
be smarter in what it does when it hits such records.
I have an update in:
http://people.freebsd.org/~gad/time-64/ac-64bit.diff
which I intend to commit sometime in the next few days. It adds
some minor sanity-checking of ut_time field, and greatly improves
the debugging messages that 'ac' produces when it is compiled
with DEBUG defined. With this change, your output would look
more like:
Doing login accounting:
root 0.84
toor 387.72
total 388.56
(Changed 5 of 59 records to have a more likely time value)
I assume it's okay to have that extra line in the summary... That
summary is only printed if bad-records were found, and I felt it
was important that we have some indication whenever these bad
record show up.
Now that I have a better idea of what the problem is, I'll try to
find what place in /usr/src might be generating these bad records.
--
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