| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Add helpers functions to query/set the offline status per backend.
Now all providers share the same offline status.
|
|
|
|
|
|
The code was still dependent on it for the ldap driver.
Changed the driver code to depend on the schema type.
Fix defaults for user and groups trees.
ATM if you use the rfc2307bis schema you have to put users and groups
in 2 separate trees (what people does by default anyway.
If this limitation will turn to be too hard, we will change this later.
|
|
The request was being freed, instead of marking it done and let the callback
free it when done. This was causing us to access freed memory, when trying to
set the next run.
Let the callback add new runs and free the request instead as normally we would
do with any other tevent_req async call.
Courtesy of valgrind again.
|
|
By attaching the reply to a subreq, we ended up freeing the operations list
element before we used it to skip to the next one.
Do not steal the context and let the unlocking code free the old reply, when it
moves onto processing the next one.
Got this one with valgrind.
|
|
After the recent changes we lost the capability to actually go offline.
Put back code that would mark the backend as offline when timeouts happen.
Make sure the enumeration code also obbeys the offline timeout, and
contributes in determining if we are offline or not.
|
|
The special persistent local database retains the original name.
All other backends now have their own cache-NAME.ldb file.
|
|
Make counter for used messages explicit.
|
|
Check if the timestamp argument of sdap_save_group_recv is NULL before
using it.
|
|
sdap_get_initgr_process() was using the wrong sdap_id_map struct
when creating the searchfilter for the initgroups() call.
|
|
New tevent library finally outlawed nested loops.
|
|
If enumerations are disabled for this domain, then do not start the
enumeration task.
|
|
The data provider backends stored a name value besides the domain
name to identify themselves to the data provider. This was the name
of the id provider. Currently the backends can have different
providers for id, authentication etc. So the name may be missleading.
Also when there are more domains with the same id provider the name
is not enough to identify the backend but the domain name is. As a
consequence the backend name is removed completely and only the
domain name is used for identification.
|
|
|
|
Currently the kerberos locator plugin needs these two variables to
be set to find a KDC which is configured in sssd but not in
/etc/krb5.conf.
|
|
|
|
Always immediately return to DP, and update users/groups in the background.
Also implements an optimization to retrieve only changed/new users/groups
by filtering using the modifyTimestamp after the first query.
|
|
|
|
First pass to remove the legacy option and make it just a property of the
provider
|
|
|
|
|
|
Try as hard as possible to store as much data as we can.
|
|
- making the realm part upper case is now optional and done in the
LDAP backend
- using a username@realm UPN is now optional
|
|
If available the original DN and the user principle will be stored
in sysdb.
|
|
If a user principle name (upn) can be found in sysdb the krb5
backend will use this otherwise is build as username@realm. It is
checked that the realm is upper case only.
|
|
|
|
Remove redundant reconnection code that was interfeering with the sbus
reconnection code.
Consolidate include files for sbus relates operations.
Make pamsrv code similar to nsssrv code.
|
|
This patch introduces provider=files as a valid provider.
Upon loading the backend, its properties in confdb are overwritten to
those that represent legacy local domain.
Also document this in sssd.conf(5) and example config
|
|
Mirrors what we have done with the monitor.
|
|
Let services identify themselves voiluntarily as the first operation
instead of polling from the monitor.
Also consolidate some common functions and make them available as monitor
helpers.
|
|
The child processes call prctl() and when their parent process is
killed, they are sent SIGTERM using prctl. This is currently
Linux-specific, for non-Linuxes, a similar effect is achieved by
catching a set of common termination signals and sending SIGTERM to the
process group.
|
|
Previously, we had hardcoded the paths for the NSS, PAM and
private PAM sockets to /var/lib/sss/pipes. With this patch, we
will specify the sockets with --with-pipe-path.
|
|
|
|
Make as much as possible static, and remove use of talloc_reference and
allocation/deallocation of memory when not necessary.
Fix also responder use of rctx->conn, was mistakenly used for both
monitor and dp connections.
|
|
This reduce code duplication as it allows to use one set of watch and timeout
functions, and at the same time also allow not to use a secondary structure just
to unify these functions.
|
|
Rationalize and rename connection names in preparatoin for merging of server and
connection structures.
|
|
Simplify code by removing stuff that is never used or redundant.
|
|
|
|
Retrieving ldap results and storing users could sometimes results in race
conditions where the final ldap result was retrieved before the store operations
where finished resulting in the operations to be aborted before termination.
Implement a serialization mechanism per operation.
|
|
|
|
Also add tests
|
|
The sdap_handle might be freed when processing a message.
Rearrange data flow so that the sdap_handle is never used after
a message is processed but a new event (dependent on the handle) is
instead scheduled. If the sdap_handle is freed, the scheduled event
is also removed and not fired
|
|
|
|
|
|
The way openldap libraries work, require to have a single engine per
connection as all replies are read at the same time. So we need to
always read anything that comes in from the wire and then loop to
dispatch results to the requests that are waiting.
|
|
|
|
|
|
|
|
|
