For reference, my smbd.conf is as follows:
# Global parameters
[global]
workgroup = SMC
realm = internal.stmaryscollege.co.uk
netbios name = XEN-ARCH-AD-01
server role = active directory domain controller
printcap name = /etc/printcap
load printers = no
dns forwarder = 192.168.0.30
#DNS handled by BIND
server services = -dns
log level = 1
max log size = 10240
log file = /var/log/samba/samba.log.%d
debug uid = yes
debug timestamp = yes
allow dns updates = secure
nsupdate command = /usr/bin/nsupdate -g
spn update command = /usr/sbin/samba_spnupdate
idmap_ldb:use rfc2307 = yes
winbind nss info = rfc2307
#Small file tuning
read raw = no
level2 oplocks = true
On 08/05/2013 10:10, Andrew Bartlett wrote:
On Tue, 2013-05-07 at 21:55 +0100, Alex Matthews wrote:
Hi there,
I have three S4 servers running as AD DCs.
In order to keep the sysvol share in sync I'm using crontab to run the
following command:
/usr/bin/rsync -PavAX --delete root@<masterPDC>:/var/lib/samba/sysvol/
/var/lib/samba/sysvol/
However everytime this command is run a couple of extra smbd processes
are started on the <masterPDC> (between 2 and 5 processes) which never
exit and just sit there taking up resources. So, quite quickly I had a
system with over 500 smbd processes and no free memory which very
abruptly fell over and stopped serving genuine clients.
Has anyone else come across this issue/know what is causing it?
I have taken some level 10 logs of the smbd processes that get formed.
However I don't have access to them from my current location. I will
email them in tomorrow from work.
My best guess is that the winbind part of samba's AD DC is making a
connection, but never closing it.
Andrew Bartlett
--
To unsubscribe from this list go to the following URL and read the
instructions: https://lists.samba.org/mailman/options/samba