Hello list, I have managed successfully to setup a fully functional openldap server on FreeBSD. So far, I had success with : ppolicy, ACLs, legacy SQL exposed as LDAP, SASL authentication. My only problem thus far is combining SASL with ppolicy. When binding with classic simple authentication using -D dn, then ppolicy overlay has the expected effect. However when using SASL (SASL/SCRAM-SHA-1) with -U, while it works correctly converting uid to DN with authz-regexp, it does not seem to look for ppolicy (default or derived from pwdPolicySubentry). Moreover, enforced violations of ppolicy (e.g. failed attempted authentications >= pwdMaxFailure) when done via SASL seem to have no effect on ppolicy attributes, e.g. pwdAccountLockedTime, while they work fine when binding with simple authentication. Is there any way to overcome this? Or is ppolicy honored only via simple DN binds?

On 10/07/2014 11:56, Howard Chu wrote: > Achilleas Mantzios wrote: >> Hello list, >> I have managed successfully to setup a fully functional openldap server on FreeBSD. >> So far, I had success with : ppolicy, ACLs, legacy SQL exposed as LDAP, SASL authentication. >> My only problem thus far is combining SASL with ppolicy. When binding with classic simple >> authentication using -D dn, then ppolicy overlay has the expected effect. >> However when using SASL (SASL/SCRAM-SHA-1) with -U, while it works correctly converting uid to DN >> with authz-regexp, it does not seem to look for ppolicy (default or derived from pwdPolicySubentry). >> Moreover, enforced violations of ppolicy (e.g. failed attempted authentications >= pwdMaxFailure) >> when done via SASL seem to have no effect on ppolicy attributes, e.g. pwdAccountLockedTime, >> while they work fine when binding with simple authentication. >> >> Is there any way to overcome this? Or is ppolicy honored only via simple DN binds? > > ppolicy is only honored by Simple Binds. There was some discussion, a long time ago, about how to make SASL use/recognize LDAP password policy, but it never went anywhere. > Thank you, I am sure there are reasons for this, if you don't bother, you might give some pointers to this discussion?

Also, is there any non-programmatic workaround for this?