Kevin Bedford
2008-Jul-03 12:38 UTC
[Samba] Samba pdc with ldap & samba3-3.0.25c-34 on CentOS 5
Hi all,
I had been running a Samba pdc using the standard tdbsam backend on
CentOS 4.5 supporting 20+ WinXPsp2 workstations and a Win2k3 Terminal
Server. This set up was running fine for arroung 18 months maybe more.
The above system was built on ageing hardware that used to be a Win 2k
server that died from OS corruption.
Due to the expanding client needs there was good reason to replace the
old box.
The new system is based on a Supermicro server board with an Intel Core2
Duo CPU E6750, 8G RAM and 4 500G HDD's as two software RAID mirrors.
This Machine was installed with CentOS 5 and Samba updated to 3.0.25c
instead of the supplied 3.0.25b as the previous setup had suffered an
inability to change passwords from Win2k3 with any prior version.
Openldap and Samba were configured as a PDC mostly from the HOWTO by
Example from samba.org and the user accounts were imported from the old
tdb backend. The intent was to migrate the domain however this did not
go as expected. All workstations needed to be removed and re joined to
the new domain and user profiles needed to be re created.
Once the new server was installed various users have experienced some
kind of random file locking related issues when working on "Word" or
"Excel" documents from the file shares, resulting in various Windows
error messages including the following
"Word failed reading from this file(~WRL0925), please restore the
network connection or replace the floppy disk and retry"
"Cannot delete filename; It is being used by another person or
program."
"Quote Register 2008.xls is locked for editing by 'drana'"
(in this
case drana is the user trying to open the file that received the error.)
"The document was saved successfully, but Excel cannot re-open it
because of a sharing violation, Please close the document and try to
open it again."
I have several other Samba file servers on CentOS 4 & 5 machines some of
which are PDC's but this is the first LDAP deployment. None of the
others have exhibited this issue.
My Research so far has found plenty of other reports of this kind of
issue with no reports of real solution. I have read some documentation
that claims OPLOCKS can cause this issue and have tried various config
options to disable them.
Other documents claim network hardware faults such as cabling or
switches and I have already replaced anything that appeared at all
suspect. Also just prior to it's retirement the old server was being
accessed using exactly the same cabling and switch infrastructure.
There also does not seem to be any issue copying large amounts of data
between servers using scp or rsync so disk and network throughput does
not seem to be an issue.
The smb.conf is as follows
[global]
unix charset = LOCALE
workgroup = EARTHLIFT
netbios name = EARTHLIFTSERVER
server string = earthliftserver
interfaces = eth0, lo
bind interfaces only = Yes
passdb backend = ldapsam:ldap://localhost
#enable privileges = Yes
#nt acl support = no
username map = /etc/samba/smbusers
log level = 1
#syslog = 0
log file = /var/log/samba/%m
max log size = 50
smb ports = 139
name resolve order = wins bcast hosts
socket options = TCP_NODELAY IPTOS_LOWDELAY SO_KEEPALIVE SO_RCVBUF=8192
SO_SNDBUF=8192
time server = Yes
printcap name = CUPS
show add printer wizard = No
add user script = /opt/IDEALX/sbin/smbldap-useradd -m "%u"
delete user script = /opt/IDEALX/sbin/smbldap-userdel "%u"
add group script = /opt/IDEALX/sbin/smbldap-groupadd -p "%g"
delete group script = /opt/IDEALX/sbin/smbldap-groupdel "%g"
add user to group script = /opt/IDEALX/sbin/smbldap-groupmod -m "%u"
"%g"
delete user from group script = /opt/IDEALX/sbin/smbldap-groupmod -x
"%u" "%g"
set primary group script = /opt/IDEALX/sbin/smbldap-usermod -g "%g"
"%u"
add machine script = /opt/IDEALX/sbin/smbldap-useradd -w "%u"
logon script = %U.bat
logon path = \\%L\profiles\%U
logon drive = H:
logon home = \\%L\%U
domain logons = Yes
preferred master = Yes
local master = yes
wins support = Yes
ldap suffix = dc=earthlift,dc=local
ldap machine suffix = ou=machines
ldap user suffix = ou=People
ldap group suffix = ou=group
ldap idmap suffix = ou=Idmap
ldap admin dn = cn=Manager,dc=earthlift,dc=local
idmap backend = ldap:ldap://localhost
idmap uid = 10000-20000
idmap gid = 10000-20000
#map acl inherit = Yes
printing = cups
hide files = /desktop.ini/Desktop.ini
os level = 65
level2 oplocks = no
kernel oplocks = no
oplocks = no
#use sendfile = no
#lock spin time = 15
[homes]
comment = Home Directories
valid users = %S
read only = No
browseable = No
[printers]
comment = SMB Print Spool
path = /var/spool/samba
guest ok = Yes
printable = Yes
browseable = No
#[profdata]
#comment = Profile Data Share
#path = /home/users/profdata
#read only = No
#profile acls = Yes
[netlogon]
comment = Network Logon Service
path = /home/users/netlogon
guest ok = yes
write list = @ntadmins
share modes = no
[Profiles]
path = /home/users/profiles
browseable = no
guest ok = yes
create mask = 0600
directory mask = 0700
writeable = yes
oplocks = false
level2 oplocks = false
csc policy = disable
veto oplock files = /prf*.tmp/;
profile acls = yes
[public]
writeable = yes
public = yes
path = /home/shares/public
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
[batch]
writeable = yes
public = yes
path = /home/shares/batch
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = yes
level2 oplocks = yes
veto oplock files = /*.mdb/*.xls/*.wdb/;
[faxes]
writeable = yes
public = yes
path = /home/shares/faxes
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
[scans]
writeable = yes
public = yes
path = /home/shares/scans
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
[admin]
writeable =yes
guest ok = no
path = /home/shares/admin
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
[excavate]
writeable =yes
guest ok = no
path = /home/shares/excavate
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = false
level2 oplocks = false
# veto oplock files = /*.mdb/*.wdb/;
[piling]
writeable =yes
guest ok = no
path = /home/shares/piling
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = false
level2 oplocks = false
# veto oplock files = /*.mdb/*.wdb/;
[ohs]
writeable =yes
guest ok = no
path = /home/shares/ohs
# force group = Domain Users
create mode = 770
directory mode = 770
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
[finance]
writeable =yes
guest ok = no
path = /home/shares/finance
# force group = finance
create mode = 770
directory mode = 770
valid users = @finance
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
[ANZ]
writeable =yes
guest ok = no
path = /home/shares/ANZ
# force group = finance
create mode = 770
directory mode = 770
browseable = no
valid users = @finance
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
[payroll]
writeable =yes
guest ok = no
path = /home/shares/payroll
# force group = payroll
create mode = 770
directory mode = 770
valid users = @payroll
oplocks = yes
level2 oplocks = yes
# veto oplock files = /*.mdb/*.xls/*.wdb/;
The longer a user has the file open for the more likely they seem to be
to have the problem.
This also occurs a lot in the log files but more frequently than the
user issues so I didn't consider it the cause of the issue
[2008/07/03 16:41:56, 0] lib/util_sock.c:read_data(534)
read_data: read failure for 4 bytes to client 192.168.0.82. Error =
No route to host
[2008/07/03 16:21:25, 0] lib/util_sock.c:read_data(534)
read_data: read failure for 4 bytes to client 192.168.0.82. Error =
Connection timed out
The only way to regain access to a file once the error occurs is to kill
the related smb process on the server.
My next step was going to be to rebuild the old server agian using
CentOS 4 to eliminate the change in kernel/OS version but retain the
LDAP server so as not to have to rebuild the domain and profiles again.
If anyone can suggest a solution or more appropriate course of action
please advise?
Best Regards
Kevin Bedford
Possibly Parallel Threads
Wisdom of the Ancients
