[db-wg] Proposed change 2003.1: notification for more-specific

On 2003-03-07 17:56:46 -0500, Larry J. Blunk wrote:
> 
> My comment is not specifically related to this proposal, but I
> have long been concerned about the optional status of the
> mnt-nfy: attribute.  This means if someone cracks a mntner
> password and begins submitting updates (which naturally will be
> successful since they have the password), no notification will
> be sent to the mntner.
> 
> I'm not sure of the best way to handle this.  I suppose there
> are some people who don't want to be bothered by notification
> of successful updates and if they're using PGP that might be
> okay.  It also seems the upd-to attribute name was very poorly
> chosen as it is difficult to differentiate it's function from
> that of mnt-nfy (I often get the two confused and I deal with
> this stuff every day).   I personally feel that the mnt-nfy
> should be mandated in RPSL and those mntner's who do not have
> one should have it replicated from their upd-to attribute
> value.  For those who really don't care to see the results of
> successful updates, they could simply direct the email address
> to /dev/null.

I am personally agnostic on the issue, but do not think there would be
a problem making "mnt-nfy:" mandatory.

As a data point, of the 8808 maintainers in the RIPE Database, there
are 2322 that use password-based authentication (CRYPT-PW or MD5-PW)
and have no "mnt-nfy:" attribute.


-- 
Shane Kerr
RIPE NCC

p.s. I also think that "upd-to:" and "mnt-nfy:" are probably not the
     best names.  But what can we expect of a standard with things
     like "mntner:" and "aggr-mtd:"?   (These names can't be to
     shorten them, or we wouldn't have "peering-set:" and
     "mbrs-by-ref:")?  Would it make sense to make an alias for
     "upd-to:" of something reasonable, e.g. "auth-fail-nfy:"?