That happens if you 'update' is rpm -e + rpm -i. It's being done
automatically - the only thing it the spec file is to create it if it
doesn't exist.
%pre
g=`cat /etc/group | grep ^ntop:`
if test ".${g}" = "."; then
/usr/sbin/groupadd -r ntop 2>/dev/null || :
fi
u=`cat /etc/passwd | grep ^ntop:`
if test ".${u}" = "."; then
/usr/sbin/useradd -s /bin/false -c "ntop server user" -g ntop \
-d %{ntoproot}/share/ntop -M -r ntop 2>/dev/null || :
fi
It can be argued that it's the RIGHT behavior for rpm -e - to unwind the
entire installation. Although it does leave the .rrd files unowned...
I use rpm -U or rpm -F all the time without problem.
-----Burton
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of
Kenneth Porter
Sent: Friday, July 25, 2003 9:23 PM
To: [EMAIL PROTECTED]
Subject: Re: [Ntop] Bug in ntop.spec
BTW, I updated from 2.1 to 2.2c and the 2.1 %postun removed the ntop user
and group, so I had to re-instate them. It might be a good idea to add
another entry to the initscript to add the user/group if missing and
initialize the password. Something like "service ntop firsttime".
_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop
_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop