On Mon, 2006-11-20 at 10:42 -0600, Jeff Vian wrote: > On Mon, 2006-11-20 at 08:31 -0500, Gene Heskett wrote: > > On Monday 20 November 2006 08:09, Anne Wilson wrote: > > >On Monday 20 November 2006 12:46, Gene Heskett wrote: > > >> Now if logrotate would work... Its not. See my other post. > > > > > >Well, for all the logs that I would expect to rotate, I have files dated > > >5.11.06, 12.11.06, 19.11.06 and 20.11.06. > > > > > >Yet another peculiarity of your install, Gene. > > > > > >Anne > > > > I *may* be on the track of it, following Ricks advice I ran it from a > > shell, and discovered the whole thing died on not finding a user 'named', > > and I do not have bind installed. So I'll do that but set the service to > > off. nscd is running and thats should be enough. > > > > As to this being a bugzilla-able item, dunno. It seems to me that it > > should be bind that installs this particular script, not logrotate's rpm. > > > > After installing bind, the user named is in the passwd file, and it runs. > > This to me is a bugzilla-able packaging problem. > > > I always have bind installed so have never seen that. > There should be logic in the logrotate script to politely ignore the > stuff that is not installed, so I would consider that worthy of a bug on > logrotate. ---- actually, there is much that he hasn't figured out. I would imagine that nscd which he thinks he needs (uncertain) might require bind-libs but wouldn't likely require bind package itself, nor should logrotate require bind. I would suspect that few fedora users actually run bind/bind-chroot and doubt that logrotate would require that any portion of bind be installed at all. trying to run logrotate scripts from /etc/logrotate.d wouldn't seem to me to possess the environmental variables or privileges that may be required and thus, seems to be a gymnastic exercise that ultimately won't prove anything - perhaps that is Gene's forte. Since logrotate runs on 100% of Fedora-6 installs out of the box, if there were a problem with logrotate, there would be MANY other user reports. Thus the problem is a local problem to Gene's computer. My suspicion is that Gene's issue with logrotate is tied to Gene's issue with cron which seems (just a wild guess here) consistent with someone who futzes with cron files, either by copying the files in from another system or by using unsuspecting text editors to edit them rather than using the only recommended method of editing via 'crontab -e' commands as the other methods are likely to cause issues with SELinux or cron itself since cron tends to reject cron files edited with uninformed text editors. Craig
