Hello there, a mildly experienced (know how to use the CLI/Terminal, edit config files and tweak easily accessible things) Linux user here.
I have used an AMD E350 Fusion Mini-ITX board to build a HTPC/Fileserver for easy access of media, mostly movies and tv-shows, running Lubuntu 12.04 with XBMC (the XBMC-XvBA variant to be precise). I have been using a guide to install and configure all of the needed software for that purpose.
Right now I have set up a basic Samba install with an unsecured global guest share for easy uploading and access to my media library from Windows clients (exclusively Windows 7 x64, one client upgraded to SP1) on a closed, wired 1Gbps network. The share is set up on a LVM stripe connecting two WD20EARS 2TB harddrives.
The problem I'm seeing is Samba being unable to maintain fast (>60MB/s) read speeds, while writes are universally fast and consistent. This is a known problem and have been remedied in the past with adding opcodes to smb.conf... but here's the fun part:
It's not consistent.
I have been restarting the service as part of debugging my problem, and have noticed that on some restarts, Samba behaves properly, with read speeds consistent with the overall network speed. Earlier on I thought it was a result of tweaking and left it to work, but recently I had to reboot the server a couple of times and the problem comes back. I have now determined that Samba behaves erratically and there is no visible pattern to whether it behaves as it should or not.
I have been trying to determine whether the problem was with the drives, the LVM share, the CPU/RAM bottlenecking the board, the network gear or the network itself and i have no clues whatsoever - local readwrite tests show absolutely no problems with accessing the data and the network works as intended. Stresstests on a "good" service start are all green, and XBMC playing a 1080p h264 file has no impact other than stressing the CPU and reading from the same drive from which the file was copied over the network. That leaves me with some potential Samba issue with which neither my current knowledge nor Google could help me. I have tried tweaking my smb.conf on my own, but other than a few optimization tweaks, "bad" service starts still occur.
Here is my current smb.conf. It is a giant mess right now owing to different methods of trying to solve my problem:
Code:
[global]
;General server settings
netbios name = SOULFORGE
server string = Domowy serwer plikow
workgroup = WORKGROUP
domain master = yes
local master = yes
preferred master = yes
load printers = no
printing = bsd
printcap name = /dev/null
disable spoolss = yes
announce version = 5.0
passdb backend = tdbsam
security = share
guest account = nobody
null passwords = true
name resolve order = bcast host lmhosts wins
wins support = yes
interfaces = lo, eth0
bind interfaces only = true
syslog = 1
syslog only = yes
log level = 1
strict locking = no
strict sync = no
sync always = no
large readwrite = no
min receivefile size = 16384
use sendfile = true
aio read size = 16384
aio write size = 16384
read raw = yes
write raw = yes
kernel oplocks = yes
max xmit = 65535
dead time = 15
getwd cache = yes
[Dane]
path = /mnt/dysk
public = yes
browseable = yes
guest only = yes
writable = yes
case sensitive = True
oplocks = yes
create mask = 0777
directory mask = 0777
Bookmarks