4.1.1.2 - Name/Password Authentication

This is the most common authentication system, though not the safest. The user provides his name and a password. Both are passed as clear text to the server, which checks that the user exists, and that its password is correct.

User’s name retrieval

The first thing the server does is to check that the user’s name exists in the server. The provided name is always a full DN.

Here is an example of simple authentication using Studio, where we authenticate the uid=admin,ou=system user :

The password is not visible here, but this is just for security reasons.

This request is sent to the server, which will check that the uid=admin,ou=system exists in its backend. If it doesn’t, the authentication will fail.

Password check

That’s not enough : once the user is retrieved, we have to check the provided password against the stored password.

The entry associated with the user should contain a userPassword AttributeType, otherwise the request will be rejected. Here is an example of such an entry :

version: 1

dn: uid=admin,ou=system
objectClass: top
objectClass: person
objectClass: organizationalPerson
objectClass: inetOrgPerson
cn: system administrator
sn: administrator
displayName: Directory Superuser
uid: admin
userPassword:: c2VjcmV0

As we can see, this entry has an userPassword which contains the base64 encoded password. If we decode the value, we get :

userPassword: secret

Not exactly safe…

Password storage

As we have just seen, the password is stored in plain text in the server. This is not exactly safe ! As soon as someone gets access to your server, all the passwords are compromised. This is certainly not the way we want to protect our users !

Hopefully, you can hash those passwords, instead of storing them as provided.

ApacheDS let you select an encryption type when you inject a password :

The following hash method are available :

How it works ?

So the server receives a Name/Password authentication request. The password is in clear text up to this point. Once the user is found in the server, and if it has a userPassword attributeType, the server extracts each values contained in this AttributeType (we may have more than one password per user) and check the provided password against those values.

This is not as simple as it seems : as we may have hashed the values on the server, we first have to detect the selected hash method, and then hash the provided password, which result is compared to the stored hashed value.

Hopefully, the hash method is stored within the hashed password in the server :

version: 1

dn: uid=admin,ou=system
objectClass: top
objectClass: person
objectClass: organizationalPerson
objectClass: inetOrgPerson
cn: system administrator
sn: administrator
displayName: Directory Superuser
uid: admin
userPassword:: c2VjcmV0
userPassword:: {CRYPT}FgGgCMynLfYGw

Here, one of the userPassword value is hashed using the crypt algorithm. The following code is used to check the provided password :

for each stored password
  if it has a hash method 
    then 
      extract the method
      hash the provided password using this method
      compare the result with the stored hash value
      if they are equal
        then
          return true
    else
      compare the provided password with the stored password
      if they are equal
        then
          return true
done

return false