John A. Sullivan III wrote:
I'm finding several weird issues with DSGW authentication which
very difficult for our users to use. Not to complain - great DS - but
we're experiencing some problems.
We do not allow anonymous browsing of the tree. Each client has a user
who has rights to search only their portion of the tree for possible
DSGW logins. The ACI, place on the root, is thus:
"uid || st || sn || ou || name || entrydn || dn || dc || objectClass ||
cn || o || l || c || givenName") (version 3.0;acl "Client DSGW
Lister";allow (search,read)(userdn =
We have an example test user named sue.sutter. The full dn is
The first step is to go the authentication page where we read:
"The first step in authenticating to the directory is identifying
This is why we created a user with rights to browse for other users and
defined it with a binddnfile entry. That part is working fine.
If I enter sue.sutter, it does not find her directly but rather offers a
list with a single hyperlinked choice. That's the first problem (a
problem for anyone with a "." in their uid). The query has replaced the
"." with a space:
filter="(&(objectClass=person)(|(sn=sue sutter)(cn=sue sutter)))
I tried surrounding it with quotes and escaping it with a back slash but
the quote was interpreted literally and the back slash gave the same
results as the period alone.
Is this a bug, a configuration error, or just the way it's supposed to
be? If the latter, this is very user unfriendly. A techie might
understand escape characters or special encoding but not an everyday
Sounds like a bug. I have no idea why a "." would be replaced with a
It wouldn't be so bad if they could simply click on the hyperlink
allowed to login. However, the hyperlink does not work. Mousing over
but it goes nowhere. A packet trace shows no packets coming from the
browser to the DS. What might we have configured incorrectly to cause
this? We see the same thing in Konqueror as we see in Firefox3 all
running on fully patched Ubuntu 8.0.4.
Sounds like another bug. Didn't you already file a bug about this issue?
Hmmm . . . this is getting long. I'll put the other problem
another email. Thanks - John