[Samba] Winbindd periodical timeout issue

[Rasmus Larsen]

Hey,

I've got an issue with winbindd seeming to periodically fail to resolve user/group names from an AD setup. I'm getting the following error, the error seems to occur when I try to access a the samba server OR a directory that uses AD dependent ACLs/permissions. The server will stall for around 30-40 seconds and then it all seems to work perfectly (though I haven't yet determined if it causes permission problems). This only seems to happen if the server or data hasn't been accessed for an extented period.

This is part of the winbindd log:

[2007/04/15 22:06:30, 1] nsswitch/winbindd.c:main(953)
  winbindd version 3.0.23c-2.el5.2 started.
  Copyright The Samba Team 2000-2004
[2007/04/16 13:13:43, 0] lib/util_sock.c:read_socket_with_timeout(497)
  read_socket_with_timeout: timeout read. read error = Connection reset by peer.
[2007/04/16 13:13:43, 0] rpc_client/cli_pipe.c:rpc_api_pipe(790)
  rpc_api_pipe: Remote machine erstorage.er.dtu.dk pipe \lsarpc fnum 0xc002returned critical error. Error was Read error: Connection reset by peer
[2007/04/16 13:13:43, 1] nsswitch/winbindd_group.c:winbindd_getgrnam(259)
  group Domain Users in domain ER does not exist
[2007/04/16 13:13:43, 1] libsmb/clientgen.c:cli_rpc_pipe_close(376)
  cli_rpc_pipe_close: cli_close failed on pipe \lsarpc, fnum 0xc002 to machine erstorage.er.dtu.dk.  Error was Read error: Success


About my setup:

I'm running samba Version 3.0.23c-2.el5.2 (CentOS 5, newest update), it's accessing an AD, of which erstorage.er.dtu.dk (as seen in the log), is a controller.

Has anyone else encountered this?