[WIP][PYTHON] Fix replPropertyMetaData sort order for schema extension attributes

[Andrew Bartlett]

Comment on the attached patches would be most welcome.  I realise that
the assertion that dbcheck no longer flaps is untested, but I have
pinned down one cause.

Details of the initial symptoms are here:

https://lists.samba.org/archive/samba/2015-June/192485.html
https://bugzilla.samba.org/show_bug.cgi?id=10973#c8

The issue is that we would in the past, before
61b978872fe86906611f64430b2608f5e7ea7ad8 sort attributeID values for
schema extension attributes with a leading 0x8 as negative numbers.

The attached patch (to git master) appears to resolve this issue (in
that my searches of replPropertyMetaData now appear correctly sorted). 

This *only* fixes the issue that prevented the replication, and should
make your server able to replicate to a new 4.2 instance with the same
properties as before. 

However, it has uncovered (serious) underlying issues. The
replPropertyMetaData now contains two metadata entiries per schema
-extended attribute, one with the high bit set (from msDS-IntID, which
flagged the issue), and one without (the original AD attributeID from
the prefixMap). 

That addresses the apparent regression (found by 4.2 being stricter,
and does so by fixing the sending DB). 

This duplication is not resolved. 

Neither is the deleted object handling that may still be blocking us
from turning on dbcheck as non-flapping in make test.

Andrew Bartlett

-- 
Andrew Bartlett
https://samba.org/~abartlet/
Authentication Developer, Samba Team         https://samba.org
Samba Development and Support, Catalyst IT   
https://catalyst.net.nz/services/samba




-------------- next part --------------
A non-text attachment was scrubbed...
Name: dbcheck-sort-rmd.patch
Type: text/x-patch
Size: 29795 bytes
Desc: not available
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20150728/0d81de6e/dbcheck-sort-rmd.bin>