> My bet is ENOENT -- and some kind of a vfs problem.  Which will stop
> my ability of fixing it right there, I'm afraid...

> |   | Do you remember what the error (from errno) usually was?
> |
> | It was always the same, that I remember, but I'm afraid the system with
> | the doctored cron was upgraded, and no longer runs that cron (the problem
> | had stopped happenening anyway).

FWIW, I had exactly same problem on NetBSD-1.3H/i386.
The errno was ENOENT as a result of stat("/var/cron/tabs", &statbuf).
(Of course, /var/cron/tabs existed on the system.)

This problem (seemed to) disappeared when cron job frequency decreased.
--
soda