Re: [ldapext] Unfinished business: password policy and VLV

[Howard Chu <hyc@highlandsun.com>]

One other point that I almost forgot - the whole issue of intruder detection 
is still rather fragile. E.g., if an attacker is running a stream of guesses 
against a user while the real user logs in, then the successful login will 
erase the current pwdFailureTime state and so allow the attacker a few more 
chances unimpeded.

I suppose this is only a small problem in general. My first response to this 
is that password failures should only be tracked ephemerally, within a 
particular DSA. I would also say they should be tracked by client IP address, 
but these days attacks by botnets makes the value of that approach less clear. 
It's also less effective in an environment using a cluster of load balanced 
DSAs. But I think it still makes more sense to track failures within a single 
DSA than to deal with replication of that state.

--
  -- Howard Chu
  CTO, Symas Corp.           http://www.symas.com
  Director, Highland Sun     http://highlandsun.com/hyc/
  Chief Architect, OpenLDAP  http://www.openldap.org/project/
_______________________________________________
Ldapext mailing list
Ldapext@ietf.org
https://www.ietf.org/mailman/listinfo/ldapext