>From a usability perspective, I like it.  But with anything run as root, I
think that the default configuration should be secure.

The accounts nobody and anonymous aren't standards, they're just
conventions.  So depending upon them to represent what they seem to say they
are isn't, IMHO, the wisest choice.

I'll agree that in the real world it's only a risk if you're already hacked.
But this change has the POTENTIAL to reduce the security of the default
configuration by some tiny amount.  Which exposes us to certain
prejudices -- which is why certain OSes are STILL distributing ports of ntop
1.1.  Think I'm joking?  See
http://www.openbsd.org/3.2_packages/i386/ntop-1.1.tgz-long.html.

All of that said, I can't think how to tighten it down other than removing
it - even if you read and parse /etc/shells and /etc/passwd and stuff, you
still can't be sure that it's an account that can't login and/or one that
doesn't have rights somewhere it shouldn't.  For example, just because a
shell has the word nologin in it's name, doesn't mean that the user can't
still login with it (rm -f /sbin/nologin; ln -s /usr/bin/bash
/sbin/nologin)!

So, since I can't fix it and you want it left in, that's that...

As to HP, well, if the setuid(-1) fails, then you're back with the old
code - you have to give a -u name parameter, so that causes me no angst.

I'm not sure which -P hack you mean... but I guess it's irrelevant anyway.

-----Burton



-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf
Of Luca Deri
Sent: Tuesday, April 01, 2003 12:32 AM
To: [EMAIL PROTECTED]
Subject: Re: [Ntop-dev] -u change (was New ntop commit (author deri))


Burton,
the idea is to type "ntop" <enter> and run the application. I think that
the -P hack is fine. If you want to change it (e.g. #ifdef HP user =
???) do that but I suggest not to remove it. My idea was not to make
ntop insecure, just to ease its use.


Cheers, Luca


Burton M. Strauss III wrote:

>We need to update ntop.8 and the generated .txt and .html files...
>
>In some implementations (HP-UX), nobody is assigned a negative uid and you
>can't setuid() to it.  I think it will fail and the code will catch it.
>Then again y'all know my issues w/ HP-UX...
>
>Also, shouldn't we really check and make sure that the logon shell doesn't
>permit login... otherwise, ghu forbid somebody hacks the nobody or
anonymous
>account???   I guess I'm uncomfortable that this opens up another reason
for
>certain 'secure' OSes to dis ntop...
>
>-----Burton
>
>
>
>
>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf
>Of [EMAIL PROTECTED]
>Sent: Monday, March 31, 2003 9:11 AM
>To: [EMAIL PROTECTED]
>Subject: [Ntop-dev] New ntop commit (author deri)
>
>
>Update of /export/home/ntop/ntop
>In directory jabber:/tmp/cvs-serv10070
>
>Modified Files:
>       globals-core.h globals-structtypes.h main.c util.c
>Log Message:
>Moved black/white list functions from the netflow plugin to util.c (this
>should
>ease their integration with other plugins such as sFlow).
>
>Changed -u behaviour. If the -u flag is *not* specified, ntop looks for
>users
>"nobody" and (if not found) for "anonymous". This way, ntop will start as
>user with no privileges without having to specify -u. If -u is specified,
>ntop
>will change user as usual.
>
>
>
>_______________________________________________
>Ntop-dev mailing list
>[EMAIL PROTECTED]
>http://listgateway.unipi.it/mailman/listinfo/ntop-dev
>
>_______________________________________________
>Ntop-dev mailing list
>[EMAIL PROTECTED]
>http://listgateway.unipi.it/mailman/listinfo/ntop-dev
>
>


--
Luca Deri <[EMAIL PROTECTED]>   http://luca.ntop.org/
Hacker: someone who loves to program and enjoys being
clever about it - Richard Stallman


_______________________________________________
Ntop-dev mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop-dev

_______________________________________________
Ntop-dev mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop-dev

Reply via email to