| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
|
|
Guenther
|
|
metze
|
|
Guenther
|
|
The previous patches to the provision system cut down on the number of
reconnects, and disabled the partition handling for part of the
process. This means we lost the setting of @OPTIONS as a replicated
attribute into the partitions.
Andrew Bartlett
|
|
This allows us to turn a python LdbMessage back into a string.
Andrew Bartlett
|
|
Also add tests to prove that we got this correct, and correct the
existing tests which used the wrong constants.
Andrew Bartlett
|
|
This changes dsdb_write_prefixes_from_schema_to_ldb() to use an
internal talloc hirarchy, so we can safely give it a NULL context from
the python.
It also fixes manual construction of the ldb_message - we now use the
right helper functions.
Andrew Bartlett
|
|
This allows us to test out the code that will do the modify of the
prefixMap, and to provide the bindings that may assist a future
upgrade script.
Andrew Bartlett
|
|
This will ensure that the GUID can be filled in correctly, and assist
us to validate DN targets in the future.
Andrew Bartlett
|
|
|
|
The aim is to create a function that is more easily wrapped for
python, so that we can write the updated prefixMap in an upgrade
script.
Andrew Bartlett
|
|
This revises tridge's commit 61ca4c491e1c13eb7d97847f743b0f540f1117c4
to use ldb_request_add_control() instead of a manual construction.
Andrew Bartlett
|
|
The objectclass module checks that the target parent exists, and
refuses renames if it doesn't exist. For this to work for deleted
objects we have to do the search in the objectclass module with the
"show deleted" control enabled.
|
|
|
|
We were failing because CN=Deleted Objects, which is marked as
'deleted' itself, could not be re-added in a reprovision.
Andrew Bartlett
|
|
Instead, use the fact that the ldb_parse_tree structure is public to
construct the 'and not deleted' clause as a structure, and apply each
filter tree to that template.
Andrew Bartlett
|
|
this test.
Jeremy.
|
|
The OIDs are not NULL terminated by the python caller, in line with
the LDB API, but we need them to be here, as we were casting them to a
string.
Andrew Bartlett
|
|
|
|
|
|
|
|
The problem is that samdb_result_string() and
ldb_msg_find_attr_as_string() both simply cast the string, rather than
ensuring the return value is NULL terminated. This may be best
regarded as a flaw in LDB, but fixing it there is going to be more
difficult.
Andrew Bartlett
|
|
metze
|
|
metze
|
|
The change to protocol version 1 was not intentional, and broke the
protocol established with the ntp.org project.
Andrew Bartlett
|
|
Put was assuming that the remote name was always absolute, and not relative to
the current remote directory.
Signed-off-by: Sam Liddicott <sam@liddicott.com>
|
|
|
|
This should ensure the debug messages do not have random characters at
their ends.
Andrew Bartlett
|
|
This is a small start on (ie, the only trivial part of) the work shown in:
http://k5wiki.kerberos.org/wiki/Projects/Samba4_Port#Samba.27s_use_of_Heimdal_symbols.2C_with_MIT_differences
(a table of all Kerberos symbols used in Samba4, and notes on where
they differ from those provided with MIT Kerberos)
Andrew Bartlett
|
|
Guenther
|
|
Added simple DRS rename support in replication. This should be done
async, and I'm not sure if we should also do any repl data updates to
indicate the rename. I'm still learning how this stuff works, but at
least this allows a rename on a DC to propogate correctly
|
|
When objects are deleted they get renamed to this container. The
container needs to exist when we provision
|
|
|
|
|
|
|
|
|
|
|
|
|
|
metze
|
|
We were getting build test 'pass' on the build farm when the provision
is failing. The cause was the use of filter pipelines in the selftest
code in the makefile. A bourne shell will by default report the error
status of the last element of the pipeline, which means when provision
failed, make thought everything had passed.
We could use the bash 'pipefail' option to fix this, but it isn't
portable, so instead I've modified the makefile to touch st/st_done
when the first element in the pipeline completes successfully. That
means we now get errors in make test, at the expense of making the
makefile even more horrible than it was.
|
|
Should finally fix bug #6136 ("groupType", "sAMAccountType" ... attributes).
|
|
the changes from Matthias didn't take account of url and lp being None
in some ldb python instances in 'make test'
|
|
- Centralise the lookups for the default domain (root) in the call "domain_dn"
- Reduce the LDB connections attempts ("connect" calls) from three to one
- tools should load faster
- Make the LDB connection init more like the "ldb_wrap_connection" call
- Load the right UTF8 casefolder which fixes up problems with special characters
(discovered by me: e.g. small "Umlaute" (ä, ö, ü, ...) in the DN weren't upcased
- so records "seemed" lost in TDB)
|
|
the call "domain_dn" from SamDB
|
|
Needed for special characters (e.g. in German "Umlaute")
|
|
right way
|
|
This reverts commit 53ef426e6f68728763436bd0cd3dd91180c00579.
As abartlet pointed out this causes to load all attributes and therefore gives
us more here than we need (only the check for the DN)!
|
|
|
|
my last patch was not even close ...
I'll leave abartlet to work out how to fix the test case
|
