IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12629
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Aug 23 03:23:55 CEST 2017 on sn-devel-144
Due to the non-fixable bug in the BUCKET macro tdbtool list printed some
other hash chainlist, not the freelist.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=12888
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The following C program demonstrates the issue:
#include <stdio.h>
#include <stdlib.h>
#include <stdarg.h>
#include <stdbool.h>
#include <string.h>
#include <unistd.h>
#include <errno.h>
#include <dirent.h>
#include <sys/types.h>
int main(int argc, char **argv)
{
int hash = -1;
int tsize_signed = 10;
unsigned int tsize_unsigned = 10;
int bucket;
#define BUCKET(hash, tsize) ((hash) % (tsize))
bucket = BUCKET(hash, tsize_unsigned);
printf("hash [%d] tsize [%d] bucket [%d]\n", hash, tsize_unsigned, bucket);
bucket = BUCKET(hash, tsize_signed);
printf("hash [%d] tsize [%d] bucket [%d]\n", hash, tsize_signed, bucket);
return 0;
}
Output:
$ ./tmp
hash [-1] tsize [10] bucket [5]
hash [-1] tsize [10] bucket [-1]
The first version is what the current BUCKET() macro does. As a result
we lock the hashtable chain in bucket 5, NOT the freelist.
-1 is sign converted to an unsigned int 4294967295 and
4294967295 % 10 = 5.
As all callers will lock the same wrong list consistently locking is
still consistent.
Stumpled across this when looking at the output of `tdbtool DB list`
which always printed some other hashchain and not the freelist.
The freelist bucket offset computation doesn't use the BUCKET macro in
freelist.c (directly or indirectly) when working on the freelist, it
just directly uses the FREELIST_TOP define, so this problem only affects
tdbtool list.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The variable stores the hashtable bucket, not the hash. No change in
behaviour.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Lumir Balhar <lbalhar@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Andrew Bartlet <abartlet@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Tue Aug 22 17:38:17 CEST 2017 on sn-devel-144
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat Aug 19 05:33:41 CEST 2017 on sn-devel-144
Also modify non-specified max_protocol to be PROTOCOL_LATEST
(currently PROTOCOL_SMB3_11).
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12881
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is all we can do with when using we allow SMB2/3 and the server supports
it, 'smb://' can't work unless we implement LLMNR and maybe WSD.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12876
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
get_ipc_connect() is only used in code paths that require cli_NetServerEnum()
to work, so it must already require SMB1 only.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12876
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12974
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Fri Aug 18 14:01:27 CEST 2017 on sn-devel-144
This avoids a lot of cpu cycles, which were wasted for each single smb
connection, even if the client didn't use kerberos.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12973
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Aug 18 10:04:57 CEST 2017 on sn-devel-144
The reason for the check is for write access as secrets.ldb is the
master database.
But secrets_fetch_or_upgrade_domain_info() just syncs the values
we got from if they got overwritten by secrets_store_machine_pw_sync().
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12973
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Add a basic test that when we use GET_ANC and the parents have linked
attributes, then we receive all the expected links and all the expected
objects by the end of the test.
This extends the test code to track what linked attributes get received
and check whether they match what's present on the DC.
Also made some minor cleanups to store the received objects/links each
time we successfully receive a GETNCChanges response (this saves the
test case having to repeat this code every time).
Note that although this test involves linked attributes, it shouldn't
exercise the GET_TGT case at all.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
This test:
- creates blocks of parent/child objects
- modifies the parents, so the child gets received first in the
replication (which means the client has to use GET_ANC)
- checks that we always receive the parent before the child (if not, it
either retries with GET_ANC, or asserts if GET_ANC is already set)
- modifies the parent objects to change their USN while the
replication is in progress
- checks that all expected objects are received by the end of the
test
I've added a repl_get_next() function to help simulate a client's
behaviour - if it encounters an object it doesn't know the parent of,
then it retries with GET_ANC.
Also added some debug to drs_base.py that developers can turn on to make
it easier to see what objects we're actually receiving in the
responses.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
This adds a new test to check that if objects are modified during a
replication, then those objects don't wind up missing from the
replication data.
Note that when this scenario occurs, samba returns the objects in a
different order to Windows. This test doesn't care what order the
replicated objects get returned in, so long as they all have been
received by the end of the test.
As part of this, I've refactored _check_replication() in drs_base.py so
it can be reused in new tests. In these cases, the objects are split up
over multiple different chunks. So asserting that the objects are returned
in a specific order makes it difficult to run the same test on both Samba
and Windows.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
Previously Samba would just drop cross-partition links where the link
target object is unknown. Instead, what we want to do is try to add the
forward link for the GUID specified. We can't add the backlink because
we don't know the target, however, dbcheck should be able to fix any
missing backlinks.
The new behaviour should now mean dbcheck will detect the problem and be
able to fix it. It's still not ideal, but it's better than dropping the
link completely.
I've updated the log so that it has higher severity and tells the user
what they need to do to fix it.
These changes now mean that the selftests now detect an error - instead
of completely dropping the serverReference, we now have a missing
backlink. I've updated the selftests to fix up any missing
serverReference backlinks before running dbcheck.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
We are going to end up supporting 2 different server schemes:
A. the old/default behaviour of sending all the linked attributes last,
at the end of the replication cycle.
B. the new/Microsoft way of sending the linked attributes interleaved
with the source/target objects.
Normally if we're talking to a server using the old scheme-A, we won't
ever use the GET_TGT flag. However, there are a couple of cases where
it can happen:
- A link to a new object was added during the replication cycle.
- An object was deleted while the replication was in progress (and
the linked attribute got queued before the object was deleted).
Talking to an Samba DC running the old scheme will just cause it to
start the replication cycle from scratch again, which is fairly
harmless. However, there is a chance that the same thing can happen
again, in which case the replication cycle will fail (because GET_TGT
was already set).
Even if we're using the new scheme (B), we could still potentially hit
this case, as we can still queue up linked attributes between requests
(group memberships can be larger than what can fit into a single
replication chunk).
If GET_TGT is set in the GetNcChanges request, then the local copy of
the target object should always be up-to-date when we process the linked
attribute. So if we still think the target object is deleted/recycled at
this point, then it's safe to ignore the linked attribute (because we
know our local copy is up-to-date). This logic matches the MS spec logic
in ProcessLinkValue().
Not failing the replication cycle may be beneficial if we're trying to
do a full-sync of a large database. Otherwise it might be time-consuming
and frustrating to repeat the sync unnecessarily.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
The server-side can potentially send the linked attribute before the
target-object. This happens on Microsoft, and will happen on Samba once
server-side GET_TGT support is added. In these cases there is a hole
where the Samba client can silently drop the linked attribute.
If the old copy of the target object was deleted/recycled, then the
client can receive the new linked attribute before it realizes the target
has now been reincarnated. It silently ignores the linked attribute,
thinking its receiving out of date information, when really it's the
client's copy of the target object that's out of date.
In this case we want to retry with the GET_TGT flag set, which will
force the updated version of the target object to be sent along with the
linked attribute. This deleted/recycled target case is the main reason
that Windows added the GET_TGT flag.
If the server sends all the links at the end, instead of along with the
source object, then this case can still be hit. If so, it will cause the
server to restart the replication from the beginning again. This is
probably preferential to silently dropping links.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
There was a bug in my previous patch where the code would verify
*all* links in the list, rather than just the ones that are new. And it
would do this for every replication chunk it received, regardless of
whether there were actually any links in that chunk.
The problem is by the time we want to verify the attributes, we don't
actually know which attributes are new. We can fix this by moving where
we store the linked attributes from the start of processing the
replication chunk to the end of processing the chunk. We can then verify
the new linked attributes at the same time we store them.
Longer-term we may want to try to apply the linked attribute at this
point. This would save looking up the source/target objects twice, but
it makes things a bit more complicated (attributes will usually apply at
this point *most* of the time, but not *all* the time).
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
Update drs_Replicate.replicate() so it handles being passed the GET_TGT
flag (more_flags). To do this, we need to always use a v10 GetNCChanges
request (v8 and v10 are essentially the same except for the more_flags).
If the replicate_chunk() call into the C bindings throws an error, check
to see whether the error could be fixed by setting the GET_TGT flag, and
re-send the request if so.
Unfortunately because WERR_DS_DRA_RECYCLED_TARGET isn't documented with
the other AD error codes, I've left it hardcoded for now (Microsoft
should be fixing up their Docs).
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
Windows replication can send the linked attribute before it sends the
source object. The MS-DRSR spec says that in this case the client should
resend the GetNCChanges request with the GET_ANC flag set. In my testing
this resolves the problem - Windows will include the source object for the
linked attribute in the same replication chunk.
This problem doesn't happen with Samba-to-Samba replication, because the
source object for the linked attribute is guaranteed to have already been
sent.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
- Update IDL comments to include Microsoft reference doc
- Add support for sending v10 GetNCChanges request (needed for the
GET_TGT flag, which is in the new 'more_flags' field)
- Update to also set the GET_TGT flag in the same place we were setting
GET_ANC (I split this logic out into a separate function).
- The state struct now needs to hold a 'more_flags' field as well (this
flag is different to the GET_ANC replica flag)
Note that using the GET_TGT when replicating from a Windows DC could be
highly inefficient. Because Samba keeps the GET_TGT flag set throughout
the replication cycle, it will basically receive a repeated object from
Windows for every single linked attribute that it receives.
I believe Windows behaviour only expects the client to set the GET_TGT
flag when it actually needs to (i.e. when it receives a target object it
doesn't know about), rather than throughout the replication cycle.
However, this approach won't work with Samba-to-Samba replication,
because when the server receives the GET_TGT flag it restarts the
replication cycle from scratch. So if we only set the GET_TGT flag when
the client encountered an unknown target then Samba-to-Samba could
potentially get into an endless replication loop.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
Currently we only check that the target object is known at the end of
the transaction (i.e. the .prepare_commit hook). It's too late at this
point to resend the request with GET_TGT. Move this processing earlier
on, after we've applied all the objects (i.e. off the .extended hook).
In reality, we need to perform the checks at both points. I've
split the common code that gets the source/target details out of the
la_entry into a helper function. It's not the greatest function ever,
but seemed to make more sense than duplicating the code.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
If the DRS client received a linked attribute that it couldn't resolve
the target for, then it would just ignore that link and keep going. That
link would then be lost forever (although a full-sync would resolve
this). Instead of silently ignoring the link, fail the transaction.
This *can* happen on Samba, but it is unusual. The target object and
linked-attribute would need to be added while a replication is still in
progress. It can also happen fairly easily when talking to a Windows DC.
There are two import exceptions to this:
1). Linked attributes that span partitions. We can never guarantee that
we will have received the target object, because it may be in a partition
we haven't replicated yet. Samba doesn't have a great way of handling
this currently, but we shouldn't fail the replication (because that breaks
basic join tests). Just skip that linked attribute and hope that a
subsequent full-sync will fix it.
(I queried Microsoft and they said resolving cross-partition linked
attributes is a implementation-specific problem to solve. GET_TGT won't
resolve it)
2). When the replication involves a subset of objects, e.g.
critical-only. In these cases, we don't increase the highwater-mark, so
it is probably not such a dire problem if we don't add the link. In the
case of critical-only, we will do a subsequent full sync which will then
add the links.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
We want to re-use this code to check that the linked attribute's target
object exists *before* we try to commit the transaction. This will allow
us to re-request the block with the GET_TGT flag set.
This splits checking the target object exists into a separate function.
Minor changes of note:
- the 'parent' argument was passed to replmd_process_linked_attribute()
as NULL, so I've just replaced where it was used in the refactored code
with NULL.
- I've tweaked the "Failed to find GUID" error message slightly to display
the attribute ID rather than the attribute name (saves repeating
lookups and/or passing extra arguments).
- Tweaked the replmd_deletion_state() logic - it only made sense to call
it in the code block where we actually found the target
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
When the DRS client encounters a linked attribute with an unknown target
object, it should return a RECYCLED_TARGET error, which should result in
the client resending the GETNCChanges request with the GET_TGT flag set.
This error code is currently documented by Microsoft under System Error
Codes (8200-8999). I contacted them and they will also add it to the
MS-ERREF doc in future.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12972
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12968
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Aug 18 04:45:03 CEST 2017 on sn-devel-144
The only record that must remain in gencache_notrans.tdb is the last_stabilize
marker. Use tdb_wipe_all and store the marker under the allrecord lock.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Thu Aug 17 15:49:00 CEST 2017 on sn-devel-144
This was legacy from times when we had just one non-transactioned gencache.tdb.
With the split into transactioned gencache.tdb and fast, non-transactioned,
mutexed clear-if-first gencache_notrans.tdb this has become unnecessary.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
This is another level of indentation, but it took me a while staring at the
if-condition to find that "locked" was assigned the result of "==0", not the
return value of tdb_nest_lock().
Best viewed with "git show -b".
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Log failure and exit if fork() or setsid() fails.
Leave the logic in the non-setsid() code as it is. This is probably
meant to fall through on failure of either opening /dev/tty or
ioctl(). Documentation for the ioctl() failure case is far from
clear.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Andreas Schneider <asn@cryptomilk.org>
Autobuild-Date(master): Thu Aug 17 11:48:32 CEST 2017 on sn-devel-144
Switch to using DBG_ERR(), wrap logging/sd_notifyf() lines.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>
no_process_group -> no_session, name -> daemon, drop _PUBLIC_.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>
Rename argument no_process_group to no_session to describe what it
actually does. Consistently use "daemon" for name of daemon argument.
Add documentation.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Andreas Schneider <asn@samba.org>
Replace the shell subunit test for script/traffic_summary.pl with a
python black box test.
This involves moving the test files to more standard locations.
Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Autobuild-User(master): Douglas Bagnall <dbagnall@samba.org>
Autobuild-Date(master): Thu Aug 17 07:59:38 CEST 2017 on sn-devel-144
Scripts to generate representative network traffic and replay this to a
samba instance. For load testing, performance profiling and capacity
planning.
traffic_learner process a file generated by traffic_summary and
generate a model that can be used by traffic_replay to
generate samba network traffic.
traffic_replay Replay a summary file generated by traffic_summary, or
use a model created by traffic_learner to generate
network traffic.
Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Pair-programmed-with: Garming Sam <garming@catalyst.net.nz>
Pair-programmed-with: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Pair-Programmed-With: Andrew Bartlett <abartlet@samba.org>
Pair-Programmed-With: Tim Beale <timbeale@catalyst.net.nz>
Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Aug 17 00:53:48 CEST 2017 on sn-devel-144
