Age | Commit message (Collapse) | Author | Files | Lines |
|
This was found based on a log provided by Ricky Nance
<ricky.nance@weaubleau.k12.mo.us>. Thanks Ricky!
In that log, over 2.5 days this particular allocation was repeated:
1715099 talloc_new: ../source4/dsdb/samdb/ldb_modules/schema_load.c:120 contains 0 bytes in 1 blocks
Andrew Bartlett
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Aug 17 06:21:18 CEST 2012 on sn-devel-104
|
|
The "serverReference" attribute is available on the "server" object
not on the "nTDSA" object.
This allows connections to RODCs, as they don't have a
E3514235-4B06-11D1-AB04-00C04FC2DCD2/${NTDSGUID}/${DNSDOMAIN}
principal.
Pair-Programmed-With: Björn Baumbach <bb@sernet.de>
metze
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Tue Aug 14 18:57:41 CEST 2012 on sn-devel-104
|
|
This uses a GUID based comparison, and avoids re-fetching the
samdb_ntds_settings_dn each time.
Andrew Bartlett
|
|
This is like samdb_reference_dn_is_our_ntdsa but without the attribute de-reference.
Andrew Bartlett
|
|
|
|
We often want to know if we own an FSMO role (for example). This tries to be more
efficient by comparing the GUID, rather than the string DN, as this does not need
to be re-fetched each time.
Andrew Bartlett
|
|
As the normal case (outside provision) uses a copy, this avoids a case
where a caller might modify a global variable accidentily.
As suggested by metze.
Andrew Bartlett
|
|
|
|
metze
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Tue Aug 14 13:58:31 CEST 2012 on sn-devel-104
|
|
Pair-Programmed-With: Björn Baumbach <bb@sernet.de>
metze
|
|
The index into the elements needs to match between
msg->elements and md->ctr.ctr1.array, which means we should
pre-allocate them with the same size.
Pair-Programmed-With: Björn Baumbach <bb@sernet.de>
metze
|
|
As this value is calculated new each time, we need to give it a context to live on.
If the value is the forced value during provision, a reference is taken.
This was responsible for the memory leak in the replication process. In the
example I was given, this DN appeared in memory 13596 times!
Andrew Bartlett
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Tue Aug 14 10:05:14 CEST 2012 on sn-devel-104
|
|
a tmp_ctx
|
|
|
|
tmp_ctx
|
|
|
|
This reworks dsdb_replicated_objects_commit() to have a proper local tmp_ctx and
to be more careful about what schema is set (only setting a global schema if
the original schema was global).
In particular, the new working_schema is not given a talloc reference
to the old schema. This ensures that the old schema can go away when
no longer used.
Andrew Bartlett
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Sat Aug 11 10:31:57 CEST 2012 on sn-devel-104
|
|
These sequence numbers were only used for telling if the schema was
changed, and are no longer directly related to the replication USN.
The per-partition replication USN can be obtained from the
@REPLCHANGED record on the per-partition database, and this is done
with an ldb_search().
Andrew Bartlett
|
|
This way we do not track both the partition seq number and the
replication USN for schema reload purposes.
We only need one indication of actual data change, and the replication
per-partition sequence number is no more expensive to obtain than the
ldb per-partition sequence number.
Andrew Bartlett
|
|
there is a libsecurity on OSF1 which clasheѕ with our security lib. see bug #9023.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Björn Jacke <bj@sernet.de>
Autobuild-Date(master): Fri Aug 10 14:22:21 CEST 2012 on sn-devel-104
|
|
This is controlled by setting write_indices_and_attributes.
Andrew Bartlett
|
|
In all callers, we must already have a attributeID for each of the
values or else we would have already given an error, or could not have
obtained the message over DRS.
Andrew Bartlett
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Thu Aug 9 11:39:54 CEST 2012 on sn-devel-104
|
|
This ensures that we do not add objects that should go into a partition, but we
simply return that an object is not present if the connection was created
before the partition was loaded. It is rare to create a new partition.
Andrew Bartlett
|
|
reloading a schema
This allows us to carefully control the loading of the schema.
Andrew Bartlett
|
|
This matches the rest of the function.
Andrew Bartlett
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Thu Aug 9 06:26:36 CEST 2012 on sn-devel-104
|
|
partition_primary_sequence_number
We always want LDB_SEQ_HIGHEST_SEQ here.
Andrew Bartlett
|
|
This simple operation does not need to be encased in generic ldb extended operations.
Andrew Bartlett
|
|
We do not need to reload the partition list to get the global sequence
number, as that number is stored in the metadata.tdb, not the ldb files.
Andrew Bartlett
|
|
metze
|
|
msg->elements on RODC replication
It's very important that the order of msg->elements and md->ctr.ctr1.array
is the same.
metze
|
|
This allows a lastKnownParent from LostAndFound to be replaced.
Andrew Bartlett
|
|
When we call ldb_module_done() rather than just calling the callback, we make log entries
that are critical in debugging.
Andrew Bartlett
|
|
Previously we would only consider renaming the local object, now we can cope with
renaming the remote object as well.
This should avoid most of the cases where Samba AD replication can just stop.
Andrew Bartlett
|
|
parent
This fixes up the lastKnownParent attribute on lostAndFound objects to have a GUID
(found by dbcheck).
Andrew Bartlett
|
|
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Sun Jul 29 16:51:34 CEST 2012 on sn-devel-104
|
|
|
|
add debugging
The logic looking for LostAndFound failed for a user, so add extensive
debugging to make this eaiser to trace down in future.
Andrew Bartlett
|
|
This should mean that a samba-tool drs replicate --full-sync forces a
replication of all objects, regardless of if we think the local name
is newer and regards any local name as being in conflict.
Andrew Bartlett
|
|
This allows us to proceed with replication when the source DC is sending us an object
which has a matching object in this NC (by name) but not by GUID.
Andrew Bartlett
|
|
Based on a patch proposal by Matthieu Patou <mat@matws.net>.
Andrew Bartlett
|
|
|
|
Normally this would be a very bad idea, but the specific case of fixing the instanceType
is the only case where this makes sense.
Andrew Bartlett
|
|
|
|
We should prevent this much further up the stack, but at least add a choke
at this point for now.
Additionally, this avoids administrator-forced replications causing
considerable damange to the directory.
Andrew Bartlett
|
|
|
|
|
|
|
|
This checks the type rather than just dereferencing the pointer.
Andrew Bartlett
|
|
|
|
This allows the parent to be renmaed while a new object is added on another replica.
This rename may also be a delete, in which case we must move it to lostandfound.
Andrew Bartlett
|