| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
When they are of the form of OIDs
|
|
It is useful to be able to control the 2 phase commit from application
code (s4 replication uses it)
|
|
When we replicate from a remote DC, we need to note the new uSN that
the local changes have resulted in, and modify the uSN that the notify
task uses to determine if it should send a ReplicaSync message back to
the remote DC. Otherwise we end up always triggering a ReplicaSync
every time we replicate from another DC
|
|
The getncchanges uSN is in our local space, so we must compare it to
the local_usn in replPropertyMetaData
|
|
we were setting local_usn after the marshall, so it wasn't going into
the object
|
|
Using DLIST_ADD_END() to construct a long list is very inefficient (it
is O(n^2). These lists are not ordered, so using DLIST_ADD() is much
better.
|
|
When doing an indexed search if we hit a corrupt record we abandoned
the indexed search and did a full search. The problem was that we
might have sent some records to the caller already, which means the
caller ended up with duplicate records. Fix this by returning a search
error if indexing returns an error and we have given any records to
the caller.
|
|
|
|
Guenther
|
|
Guenther
|
|
Guenther
|
|
Guenther
|
|
In time, this should avoid the astounding (order) complexity of the
objectclass sorting in objectclass.c eventually.
Andrew Bartlett
|
|
This starts the process to have Samba4 use lmhosts.
Andrew Bartlett
|
|
metze
|
|
We don't do well (even just trying to create duplicate
servicePrincipalName values) with some of these combinations, so kill
it off early before the administrator thinks it's going to work.
Andrew Bartlett
|
|
This script can be used to upgrade a provision that didn't integrate extended dn.
It can also be used to add missing extended DN that weren't created during provision.
|
|
Ensure we don't regress.
Jeremy.
|
|
On a rename we need to update uSNChanged, and the max uSN for the
partition
|
|
a rename may have no attribute changes
|
|
We could get a double free with multiple linked attributes in a
message
|
|
Windows does not seem to be always setting up repsTo using
DsUpdateRefs(). For now we will fall back to using repsFrom if repsTo
is empty. This is almost certainly incorrect, but it does get
notification based replication working with both w2k3 and w2k8.
|
|
(#6722)
|
|
Michael
|
|
We need to create the header file before we generate the source file.
metze
|
|
The issue was that we referenced an asn1compile generated file by
name, not by the subsystem is was generated by.
Andrew Bartlett
|
|
This new option avoids a lot of code that was used to intuit what
files would be output by the compiler.
Andrew Bartlett
|
|
The dreplsrv_notify code checks the partition uSN values every N
seconds, and if one has changed then it sends a DsReplicaSync to all
the replication partners listed in the repsTo attribute for the
partition.
|
|
At least on the command line the braces are needed. Strange.
|
|
|
|
These are used to load/save the per-partition uSN values managed by
the repl_meta_data module
|
|
You can now attach a partition control to searches to search within a
specific partition. This is used to get at the per-partition
@REPLCHANGED object
|
|
This object tracks the highest uSN in each partition. It will be used
to allow us to efficiently detect changes in a partition for sending
DsReplicaSync messages to our replication partners.
|
|
Guenther
|
|
Guenther
|
|
This patch brings up those subcontainers and fixes up the "systemFlags" on the
"Subnet" entry.
|
|
|
|
The directory names (MACHINE, USER) are upcased to help locating the default
group policies under the SYSVOL dir (the additional ones have only the first
letter upcased of those directory names).
|
|
metze
|
|
|
|
The partition module normally makes the sequence number extended op
operate across all partitions. It will be useful in the repl task to
be able to ask for the sequence number of one partition
|
|
It's useful seeing the object count without as much detail
|
|
These memory leaks were mostly caused by the fact that
refresh_partitions is now called periodically
|
|
Without this the client will not update its repsFrom highest_usn values
|
|
When the client tells us the highest_usn they have is N, then we want
to send them objects with usn>N, not>=N, as otherwise we end up
sending them the same object (the one with the highest uSN) again and
again.
|
|
uSNChanged>=N is good enough, and offers a possibility of a simple
optimisation where the partition module could look for that expression
and check the partitions sequence number, then avoid searching a
partition that doesn't have any records with a larger uSN.
|
|
we only need to allocate a new sequence number when
replPropertyMetaData is changing or being created on an object
|
|
I think these modules ended up LGPL because someone based the module
on an existing LGPL module in the core ldb, and it spread from
there. Certainly there is no reason for the ldb modules that are not
distributed as part of ldb to be LGPL.
|
|
We were relying on the uSNChanged>=n search always finding the DN of
the root of the partition, but this now doesn't happen very often as
we are now restricting when we change uSNChanged. This means we need
to always load the replUpToDateVector attribute from the NC root and
use it to populate the cursors in the return.
|
|
We now have dsdb_loadreps() and dsdb_savereps()
|
