FreeBSD 10.1 amd64
While trying to update Samba3+LDAP domain to Samba4 AD domain got this error:
Sorting rpmd with attid exception 3 rDN=CN DN=CN=bilyak.i,CN=Users,DC=sdr,DC=tld
convert_string_talloc: Conversion error: Illegal multibyte sequence(▒рина Сергеевна)
Conversion error: Illegal multibyte sequence(▒рина Сергеевна)
Failed to modify account record CN=bilyak.i,CN=Users,DC=sdr,DC=tld to set user attributes: objectclass_attrs: attribute 'displayName' on entry 'CN=bilyak.i,CN=Users,DC=sdr,DC=tld' contains at least one invalid value!
There is a problem in converting russian capital "И" character. Appeared at any field where this character appear.
Any updates about this issue?
You need to confirm that 'unix charset = utf8' and that your displayName is UTF8 in the existing LDAP server.
I have unix charset = utf-8 in my smb.conf
Server locale is
And records in LDAP also in utf-8.
Just tried migration on FreeBSD 11 p8 with samba 4.5.7 from ports.
Issue still persist - http://i.imgur.com/AkZUDlO.png
Workaround is to temporary change russian И in all records to something else.
if "locale" is utf-8 is not relevant here. Also not utf-8 encoding in the LDAP previous server. Only the "unix charset" setting of smb.conf that was used during classicupgrade matters and I'm quit sure that this was not UTF-8.
$ echo -n '▒рина Сергеевна' | hd
00000000 e2 96 92 d1 80 d0 b8 d0 bd d0 b0 20 d0 a1 d0 b5 |........... ....|
00000010 d1 80 d0 b3 d0 b5 d0 b5 d0 b2 d0 bd d0 b0 |..............|
The first three bytes *are* valid utf-8, but they are the bytes for '▒', not 'И'. C.f. https://www.mclean.net.nz/ucf/?c=U+2592
$ echo -n 'И' | hd
00000000 d0 98 |..|
Presumably '▒' is substituting for the invalid bytes, but the mystery is how a single character got to be invalid while the rest is clearly utf-8.