Hi,

Has anybody deployed a CTDB cluster(http://ctdb.samba.org/) successfully
with multiple nodes serving SMB simultaneously authenticating via Active
Directory(using Winbind)? If yes, can you please share /etc/samba /smb.conf and /etc/sysconfig/ctdb?

I have set up a 2-node CTDB cluster serving NFS and CIFS authenticating Windows and Linux users via Active Directory.

CTDB is all up and fine. However, the Windows client node is able to mount/access only one of the CTDB managed SMB servers. When you try to network mount from the other CTDB SMB server, the session just gets terminated. Following is the snippet of log.smb.client.

[snip]
[2008/12/21 12:29:35, 10] smbd/share_access.c:user_ok_token(231)
user_ok_token: share global-share is ok for unix user TESTDOMAIN+peyton
[2008/12/21 12:29:35, 10] smbd/share_access.c:is_share_read_only_for_token(273)is_share_read_only_for_user: share global-share is read-write for unix user TESTDOMAIN+peyton
[2008/12/21 12:29:35, 1] smbd/service.c:make_connection_snum(1190) d2950-12 (::ffff:192.168.97.1) connect to service global-share initially as user TESTDOMAIN+peyton (uid=10778326, gid=10777729) (pid 28873)
..
..
[2008/12/21 12:29:35, 3] smbd/reply.c:reply_tcon_and_X(727)
tconX service=GLOBAL-SHARE
[2008/12/21 12:29:35, 10] lib/util.c:dump_data(2223)
[000] 41 3A 00 4E 00 54 00 46 00 53 00 00 00 A:.N.T.F .S...
[2008/12/21 12:29:35, 5] lib/util_sock.c:read_socket_with_timeout(928)
read_socket_with_timeout: blocking read. EOF from client.
[2008/12/21 12:29:35, 10] smbd/process.c:receive_smb_raw_talloc(276) receive_smb_raw: NT_STATUS_END_OF_FILE
[2008/12/21 12:29:35, 3] smbd/process.c:smbd_process(2035) receive_message_or_smb failed: NT_STATUS_END_OF_FILE, exiting
[/snip]

However, if I restart Winbind on the SMB server (the client was initially UNABLE to mount from), mount works. BUT mount/access to the first SMB server fails. So, at a given instance Windows client are able to mount only from one of the CTDB managed SMB servers. I snooped network packets on port 445 + CTDB/Samba/Winbind logs but could not find major error message then the above.

From a Linux smbclient, I can mount/access from both of the SMB server.

I guess, Iam missing something important. Please advise.

Thanks in Advance,
-Tim

I presume your setup is a HA setup. I understand ctdb utilies IP takeover scheme, in case the current ctdb node fails. In your case, why would you want to connect to both nodes at the same time? The ctdb cluster is presented as a single virtual host to the clients and let ctdb manage how you connect to the individual nodes.