I agree, however, if we were to implement that at the RSP level, wouldn't
they be able to go to another RSP,
and ask for the password, and get it? I have a password utility that will
send the password to my customers, but
It won't work for one that is not under my RSP username. I don't know if the
utility that you use, is the same, since
I have not yet tried it, so I can't really comment on that at the RSP level
until I'm informed how it works in the manage.cgi
to send it, or until I finish integrating it into our system.

Thanks
Richard.
http://register.firstratehosting.com/cgi-bin/reg_system.cgi


----- Original Message -----
From: "Scott Allan" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, January 23, 2002 4:00 PM
Subject: Re: Some improvements we would like feedback on....


> I like your security suggestions - I have had similar thoughts myself
> often. I am pretty sure we could not force this on everyone (or anyone for
> that matter...). This could be implemented at the RSP level, and be a
> distinctive selling point. We could also implement an optional opt-in
> "challenge" system, which is a decent idea. Would people pay extra for it?
> (not that we would *have* to charge for it).
>
> The little I know about security:
>
> - no system is perfect
> - higher security generally means lower usability
> - you want to generally engineer your systems one or two iterations ahead
> of "easily" exploitable, generally trying to require more value of effort
> to "steal" than the value of any asset
> - there are external factors, like how easily exploitable other similar
> targets are
> -  one of the best metrics for any system is the amount of abuse it
> attracts, currently we are well within what I consider an "acceptable"
> level, certainly we need to make sure we stay there and continue to evolve
> our security options
>
> Thanks for your feedback  though - great stuff!
>
> Regards,
>
> sA
>
> At 03:23 PM 1/22/02 -0600, POWERHOUSE wrote:
> >Hello everyone,
> >Pop3 Email passwords are crackable, If a cracker is able to crack a pop3
> >email password, all that one has to so
> >is setup the profile in Outlook Express, to download all the email, and
> >leave it on the server too. That is a more
> >"detectable" way of doing it, but if all they wanted where to set up that
> >email, Then have the password sent to them,
> >check the email, then hijack the domain, then delete the profile from
> >outlook express, They could even delete that email
> >from the server by checking the box in Outlook Express to delete the
> >messages off the server when deleting from the computer.
> >If they only deleted that ONE email, it would then delete it from
existence
> >on the server, then they could delete the profile from
> >Outlook Express, and the domain owner would never know it, until their
> >domain was not working. They would not receive their
> >password, so they would not even be suspicious, and if it was the same
> >domain name, that their email is coming from, they will
> >know it when they no longer get their email, and just get a error
message.
> >
> >I recommend this instead. I would start with ALL NEW customers, and have
> >them fill out a 3 part question during
> >registration. First question, should be a Question and Answer. Stating to
> >them, that it should be a question they will
> >NEVER forget, like maybe their first animal as a child, if they remember
it
> >now, they probably will later too.
> >The second question should be what their D.O.B. is in MM/DD/YYYY format.
> >And finally the third question should be their city of birth, in all
> >lowercase, in case they can't remember how they typed
> >it in when they need their password.
> >
> >Now, in the database where the password is stored for the domain
management,
> >if they forget it, they should NOT get it
> >in the mail, instead, they should have to answer all 3 of those
questions,
> >and then it reset it for them. Now if a cracker
> >was to compromise the domain owners computer, its highly UNLIKELY that he
> >will find ALL 3 ANSWERS on the
> >domain owners computer, and not likely that he'll find them sniffing
emails,
> >or downloading emails.
> >
> >That being for all new registrations. For current customers, they should
get
> >prompted to answer those questions the first
> >time they login to manage their domain, HOPEFULLY, it won't be a cracker
who
> >compromised the account, and is signing
> >in, and then ends up filling in the information. Of course if that
happens,
> >the cracker is going to hijack the domain ANYWAYS,
> >so they can all be fixed at the same time.
> >
> >All that you'd have to do for NEW registrations is add those hidden
fields
> >to the Database, then add it to the required information
> >to register the domain name. That would NOT be hard at all. The hardest
part
> >would be if they need their password, and you have
> >to make the interface ask them all 3 questions, THEN, if they area all
right
> >prompt them for a new password, and change it at that point.
> >Then send an update to ALL RSP's telling them their is new REQUIRED
> >information for security to new registrations. FORCE all of them
> >abide by the new security measures by adding the extra few lines of code
> >into the setup profile template, and into the reg_system.cgi or
> >equivalent file, that would support this issue.
> >
> >I'm sure that 90% of all clients would appreciate this security, the
other
> >10% would probably be those who don't worry about security,
> >and are in to big of a hurry to fill out the information.
> >
> >That is how my company would do it. Security is always a number 1
priority,
> >and should be, I know a few crackers, and they are very
> >malicious. They love breaking into things, they get a high off of it.
> >
> >Richard Jones
> >[EMAIL PROTECTED]
> >http://register.firstratehosting.com/cgi-bin/reg_system.cgi
> >
> >
> >----- Original Message -----
> >From: "Paul Chvostek" <[EMAIL PROTECTED]>
> >To: <[EMAIL PROTECTED]>
> >Sent: Tuesday, January 22, 2002 9:53 AM
> >Subject: Re: Some improvements we would like feedback on....
> >
> >
> > >
> > > On Tue, Jan 22, 2002 at 08:22:46AM -0500, Scott Allan wrote:
> > > > >
> > > > >Although I realize there are issues here, it would be nicer to
allow me
> >to
> > > > >query for a users password using the (secure) RWI and then I can
> > > > >communicate that password to my customer securely (notably with a
PGP
> > > > >encrypted message).
> > > >
> > > > I guess my response would be that should someone's email account
become
> > > > compromised (or data sniffed), the ability to do all sorts of damage
has
> > > > always been there. I am not sure how to design against this -
allowing
> > > > registrants to have their U:P combo sent to them is a really useful
> > > > feature, and is pretty standard. I can't think of a way that
improves
> > > > security without seriously compromising usability... PGP is nowhere
near
> > > > widely enough deployed - I guess we could let resellers globally
disable
> > > > this for their names, but that would likely not be an option that
many
> > > > would choose, therefore not greatly improving security (it would of
> >course
> > > > allow those who desire greater security to have it).
> > >
> > > I would be heartily interested in using a system which allowed the
> > > password to be collected via the API for secure dissemination to the
> > > customer by the RSP.  We already give customers the option of
receiving
> > > their invoices via email that is PGP signed, encoded or both.  It has
> > > been a silent beef of mine for some time that the only way to get a
> > > customer his domain management password is in cleartext email.
> > >
> > > > My understanding (perhaps wrong) is that plain text data (password)
> > > > sniffing exploits are pretty rare - anyone violently disagree? It
has
> > > > always struck me as something that it is possible, but not generally
> >worth
> > > > it. In this case, not only would you have to be able to guarantee
you
> >could
> > > > get all the mail sniffed, but also be familiar with the OSRS manage
> >system.
> > >
> > > It really depends on the particular RSP.  We had an incident just a
few
> > > weeks ago in which a unix server close to a router at our upstream was
> > > compromised, and the leftover logs that were discovered indicated that
> > > cleartext POP3 logins to my co-lo customers had been properly sniffed.
> > > It's unknown how long the cracker was sniffing, and/or if he managed
to
> > > grab FTP, IMAP, HTTP and telnet passwords that simply didn't have
> > > leftover logs on the box.  In November, a customer's Cobalt on my
> > > network was compromised, and the cracker may have managed to snarf a
> > > few days worth of POP3 logins from my customer's dialup pool.
> > >
> > > These things *do* happen, in some places more frequently than in
others.
> > > If the development tools to protect myself from them were available, I
> > > would certainly use them.  Does OpenSRS feel there's a liability issue
> > > in providing cleartext passwords to RSPs?
> > >
> > > --
> > >   Paul Chvostek
<[EMAIL PROTECTED]>
> > >   Operations / Development / Abuse / Whatever       vox: +1 416
598-0000
> > >   it.canada
http://www.it.ca/
> > >
> > >
> > >
>
> Scott Allan
> Director OpenSRS
> [EMAIL PROTECTED]
>
>
>

Reply via email to