I am building Samba 4.6.6 in AIX 6.1 machine using xlc compiler. Configure and make is successful. But while doing "make check", it fails with "account [force_user] already exists in /samba/samba-4.6.6/st/nt4_dc/private/passwd" error. I am not seeing this error in linux machine. Below are the configuration options i have used. ./configure --without-ntvfs-fileserver --enable-selftest --without-ad-dc make make install .... .... [42(0)/2024 at 0s] samba.tests.kcc.ldif_import_export [43(0)/2024 at 0s] wafsamba.duplicate_symbols tdbsam_open: Converting version 0.0 database to version 4.0. tdbsam_convert_backup: updated /samba/samba-4.6.6/st/nt4_dc/private/passdb.tdb file. account[force_user] already exists in '/samba/samba-4.6.6/st/nt4_dc/private/passwd' at /samba/samba-4.6.6/lib/nss_wrapper/nss_wrapper.pl line 404. _samr_create_user: Running the command `/samba/samba-4.6.6/lib/nss_wrapper/nss_wrapper.pl --passwd_path /samba/samba-4.6.6/st/nt4_dc/private/passwd --type passwd --action add --name force_user --gid 65533' gave 2 Failed to add entry for user force_user. Unable to set password for force_user account UID_WRAPPER_ROOT=1 ./bin/smbpasswd -c /samba/samba-4.6.6/st/nt4_dc/lib/server.conf -L -s -a force_user > /dev/null at /samba/samba-4.6.6/selftest/target/Samba3.pm line 1260. Unable to create force_user at /samba/samba-4.6.6/selftest/target/Samba3.pm line 2087. ALL OK (0 tests in 0 testsuites)
It's actually "Make check" that fails. I mistakenly put "make install"
without a AIX machine that is availble for developers from the samba team to log in to, there is little chance that such issues can be fixed. As you seem to work for the manufacturer of the machine you might be able to organize a machine sponsoring for the samba team though ...
That makes real sense. There is this GCC compile farm project where one can request for an account for accessing various machines. They also have AIX 7.1/7.2 machines. One has to list the contributions for the project to get access. So i am thinking you guys requesting for an account there (rather than myself ) will make it forward.
This is the link https://gcc.gnu.org/wiki/CompileFarm
Björn, is this still an issue?
I have no idea. Honestly I think our torture/test stuff is so focused on Linux only that it will be a huge bunch of work to get that work. Unfortunately portability has very low priority these days. For that reason I would propose to close this bug as wontfix actually.
(In reply to Björn Jacke from comment #9) I agree. We don't have the resources to persue this right now, and if we did we would start with FreeBSD or similar within GitLab CI. Sorry!