Samba 4.1.6 winbind failing to start

Question

I created a clean install of Arch on the 17th Feb with Samba 4.1.4 as an "active directory domain controller" and provisioned my domain. It was working as expected providing kerberos tickets, shares, DNS, and a few GPOs. I upgraded Samba to 4.1.5 on 4th March and it was still fine. On the 19th March I upgraded Samba to 4.1.6.
About the same time Samba began failing on startup. Unfortunately my earliest journalctl entry is 20th March so I can't quite be sure that it was the upgrade which caused the failure; I've also been playing about with pam_mount. My problem is that I can't find any information on this particular error and I'm not sure where else to look. I've pasted the error message below. It seems winbind won't start but I can't figure out why.
samba[285]: samba version 4.1.6 started.
systemd[1]: Started Samba AD Daemon.
samba[286]: [2014/03/20 21:54:19.100234, 0] ../source4/smbd/server.c:492(binary_smbd_main)
samba[286]: samba: using 'standard' process model
samba[286]: [2014/03/20 21:54:19.116899, 0] ../source4/smbd/service.c:98(server_service_startup)
samba[286]: Failed to start service 'winbind' - NT_STATUS_INVALID_SYSTEM_SERVICE
samba[286]: [2014/03/20 21:54:19.117134, 0] ../source4/smbd/server.c:497(binary_smbd_main)
samba[286]: Starting Services failed - NT_STATUS_INVALID_SYSTEM_SERVICE
systemd[1]: samba.service: main process exited, code=exited, status=1/FAILURE
systemd[1]: Unit samba.service entered failed state.
And this is the output of testparm.
Load smb config files from /etc/samba/smb.conf
rlimit_max: increasing rlimit_max (1024) to minimum Windows limit (16384)
Processing section "[netlogon]"
Processing section "[sysvol]"
Processing section "[private]"
Processing section "[public]"
Processing section "[media]"
Loaded services file OK.
Server role: ROLE_ACTIVE_DIRECTORY_DC
Press enter to see a dump of your service definitions
[global]
workgroup = XXX
realm = XXX.EXAMPLE.COM
server role = active directory domain controller
passdb backend = samba_dsdb
kerberos method = secrets and keytab
template shell = /bin/bash
winbind enum users = Yes
winbind enum groups = Yes
winbind use default domain = Yes
winbind refresh tickets = Yes
server services = s3fs, rpc, nbt, wrepl, ldap, cldap, kdc, drepl, winbind, ntp_signd, kcc, dnsupdate
rpc_server:tcpip = no
rpc_daemon:spoolssd = embedded
rpc_server:spoolss = embedded
rpc_server:winreg = embedded
rpc_server:ntsvcs = embedded
rpc_server:eventlog = embedded
rpc_server:srvsvc = embedded
rpc_server:svcctl = embedded
rpc_server:default = external
idmap_ldb:use rfc2307 = yes
idmap config * : backend = tdb
acl group control = Yes
inherit acls = Yes
map acl inherit = Yes
map archive = No
map readonly = no
store dos attributes = Yes
vfs objects = dfs_samba4, acl_xattr
[netlogon]
path = /var/lib/samba/sysvol/xxx.example.com/scripts
read only = No
[sysvol]
path = /var/lib/samba/sysvol
read only = No
[private]
path = /export/home/%D/%U
[public]
path = /export/public
[media]
path = /export/media
Does anyone have any suggestions on debugging steps? Preferably without losing the contents of the AD.

Answer 1

I had exactly the same issue after rebooting my samba DC yetserday. I checked pacman.log for what had changed since last sucessfull startup and correlated my issue to this upgrade:
/var/log/pacman.log wrote:[2014-03-21 18:40] [PACMAN] upgraded talloc (2.0.8-2 -> 2.1.0-1)
[2014-03-21 18:40] [PACMAN] upgraded tevent (0.9.19-1 -> 0.9.21-2)
I've found two workarounds for this issue:
downgrading these two packages back to their old version
BUG #39557: adding /usr/lib/tevent to a newly created /etc/ld.so.conf.d/tevent.conf file and running ldconfig.
Last edited by lianse (2014-03-22 10:05:41)