Age | Commit message (Collapse) | Author | Files | Lines |
|
https://fedorahosted.org/sssd/ticket/1008
|
|
We were polluting the primary Python space with several
dependencies. We will now install them their own directory/module.
|
|
|
|
|
|
This function is no longer necessary because we don't have fake user
entries any more. The original purpose of this function was to check if
there are fake user entries for particular user and, if yes, to update
its membership.
|
|
Since there are two attributes storing information about user
memberships of the group we have to include both of them in results.
This will apply only for objects that have ghost members (i.e. they
contain the SYSDB_GHOST attribute). If an object has this attribute,
values of this attribute are not projected to the memberuid attribute.
|
|
It is remotely possible to have sysdb in an inconsistent state that
might need upgrade. Consider scenario when user asks for group
information. Some fake users are added as a part of this operation.
Before users can be fully resolved and stored properly, SSSD is shut
down and upgrade is performed.
In this case we need to go over all fake user records (uidNumber=0) and
replace each of them with ghost record in all group objects that are stated in
its memberof attribute.
|
|
|
|
Deleted sysdb_add_fake_user():
This function is no longer used.
Modified sysdb_add_user():
When user object is added to sysdb, it is important to iterate over all
groups that might have its name or any of its aliases as ghost member
and replace this ghost membership by a real one. This will eliminate
duplicite memberships.
|
|
|
|
The original approach was to store name and original DN in an object in
sysdb. When later referenced as member of a group, it was retrieved by
its original DN and the correct information about its sysdb DN was
stored in the group object which referenced it.
The new approach doesn't use fake user objects, therefore this
information has to be reached differently when constructing group
memberships. The approach is to store all users to a hash table where
original DN is used as the key and username as value. When constructing
group memberships, the name is retrieved from this hash table instead of
sysdb. This hash table is constructed when retrieving user objects from
LDAP server - if the user is not present in sysdb, it is automatically
stored in the hash table.
Another situation is for rfc2307. Because there is no nesting there, we
can construct the SYSDB_GHOST attribute directly and therefore don't
need a hash table of ghost users.
|
|
|
|
|
|
When fixing an endianness bug, we changed the protocol unnecessarily.
|
|
https://fedorahosted.org/sssd/ticket/1348
|
|
|
|
There was an issue when IPA provider didn't set PAM_SUCCESS when
successfully finished loading SELinux user maps. This lead to the map
not being read in the responder.
|
|
structure
https://fedorahosted.org/sssd/ticket/1343
|
|
|
|
|
|
https://fedorahosted.org/sssd/ticket/1330
|
|
|
|
https://fedorahosted.org/sssd/ticket/1336
|
|
The fact that we were keeping it in memory for the full duration
of the cache timeout meant that we would never reap the benefits
of the midpoint cache refresh.
https://fedorahosted.org/sssd/ticket/1340
|
|
|
|
|
|
|
|
To avoid conflicts with struct dom_sid used by samba the sss_ prefix is
added to the struct used by libsss_idmap.
|
|
|
|
|
|
|
|
|
|
Fixes a regression in the local domain tools where sss_groupadd no longer
detected a GID duplicate. The check for EEXIST is moved one level up into
more high level function.
The patch also adds the same rename support for users. I found it odd that
we allowed a rename of groups but not users. There is a catch when storing
a user -- his cached password would be gone. I think that renaming a user
is such a rare operation that it's not severe, plus there is a warning in
the logs.
|
|
documentation-access_provider
Signed-off-by: Stephen Gallagher <sgallagh@redhat.com>
|
|
|
|
https://fedorahosted.org/sssd/ticket/949
|
|
The subdomains back end request was sending replies in a format the
responder did not understand in case the request failed.
|
|
https://fedorahosted.org/sssd/ticket/1329
|
|
Active Directory 2008R2 allows only 1500 group members to be
retrieved in a single lookup. However, when we hit such a
situation, we can take advantage of the ASQ lookups, which are not
similarly limited.
With this patch, we will add any members found by ASQ that were
not found by the initial lookup so we will end with a complete
group listing.
https://fedorahosted.org/sssd/ticket/783
|
|
This is necessary because in several places in the code, we are
appending to the attrs returned from this value, and if we relied
on the map size macro, we would be appending after the NULL
terminator if one or more attributes were defined as NULL.
|
|
|
|
This option will allow administrators to set a default shell to be
used if a user does not have one set in the identity provider.
https://fedorahosted.org/sssd/ticket/1289
|
|
This option is similar to override_homedir, except that it will
take effect only for users that do not have an explicit home
directory specified in LDAP.
https://fedorahosted.org/sssd/ticket/1250
|
|
When the ldap child process is killed after a timeout, try the next KDC.
When none of the ldap child processes succeed, just abort the connection
because we wouldn't be able to authenticate to the LDAP server anyway.
https://fedorahosted.org/sssd/ticket/1324
|
|
* Previously only the side effect was described.
|
|
https://fedorahosted.org/sssd/ticket/1316
|
|
|
|
Previous version of the SSSD did not abort the async LDAP search
operation on errors. In cases where the request ended in progress, such
as when the paging was very strictly limited, the old versions at least
returned partial data.
This patch special-cases the LDAP_SIZELIMIT_EXCEEDED error to avoid a
user-visible regression.
https://fedorahosted.org/sssd/ticket/1322
|
|
https://fedorahosted.org/sssd/ticket/1325
|
|
|