Terrible intranet performance

This forum is for Marvell Kirkwood devices such as the GoFlex Home/Net, PogoPlug v1/v2, SheevaPlug, and ZyXEL devices.

Terrible intranet performance

Postby Ophidian » Mon Jan 14, 2013 12:59 am

Hey all!

I'm relatively new to administering a Linux system, but using guides and forum topics I've managed to get a GoFlex Home 3TB set up pretty well. I have a Hybrid MBR and working FTP and SAMBA servers.

The issue is... when I try to transfer files using either FTP or SAMBA, I get 20-30 KB/sec. I'm really confused about this. I tried using wget on a large internet hosted file, and got about 10Mbit. *edit* I'm on a 100Mbit Ethernet connection. I've connected it straight to the win 8 box which has Gbit Ethernet, but the same results. Prior to modding the GFH, I used samba and got 50 MB/sec.

Any thoughts? I realize I may have to come back with some config info. I thank you for your patience... I'm so close to having a good media server going.

Cheers!
Ophidian
 
Posts: 5
Joined: Mon Jan 14, 2013 12:55 am

Re: Terrible intranet performance

Postby moonman » Mon Jan 14, 2013 1:38 am

run top/htop while transferring data to see if CPU is busy with something else. what filesystem did you format the partition where you are transferring the data from/to. Would be nice to see your smb.conf as well.
Pogoplug V4 | GoFlex Home | Raspberry Pi 4 4GB | CuBox-i4 Pro | ClearFog | BeagleBone Black | Odroid U2 | Odroid C1 | Odroid XU4
-----------------------------------------------------------------------------------------------------------------------
[armv5] Updated U-Boot | [armv5] NAND Rescue System
moonman
Developer
 
Posts: 3388
Joined: Sat Jan 15, 2011 3:36 am

Re: Terrible intranet performance

Postby WarheadsSE » Mon Jan 14, 2013 1:39 am

$this->bbcode_second_pass_quote('Ophidian', 'T')he issue is... when I try to transfer files using either FTP or SAMBA, I get 20-30 KB/sec.

Holy crap, that's really sucky. You should be able to cap out at upwards of 30MB/s with samba to ext3/4, and nearly 50MB/s over FTP. Keep an ssh session open running htop (pacman -Sy htop) and see what's up with the performance.
Core Developer
Remember: Arch Linux ARM is entirely community donation supported!
WarheadsSE
Developer
 
Posts: 6807
Joined: Mon Oct 18, 2010 2:12 pm

Re: Terrible intranet performance

Postby Ophidian » Mon Jan 14, 2013 2:09 am

Thank you very much for the fast response. I loaded htop, and ran it while transferring a large file. Curious, I'm get 470-500KiB/sec now. Still short of what I should get though. As I watch htop, the cpu useage for proftpd spikes and goes low frequently. I don't know if this is normal. *edit* Oops, I forgot to mention the file systems. I have 3 partitions on the drive, 20 gig ext3 for ALARM, 1 gig swap, and remainder as EXT4. I'm transferring from a Win8 box using NTFS.

From htop:
$this->bbcode_second_pass_code('', '
CPU[|| 2.0%] Tasks: 21, 0 thr; 1 running
Mem[|||||||||||||||||||||||||||||||||||||||||||24/120MB] Load average: 0.13 0.04 0.05
Swp[| 0/1023MB] Uptime: 01:28:40

PID USER PRI NI VIRT RES SHR S CPU% MEM% TIME+ Command
376 root 20 0 5580 1896 1192 R 2.0 1.5 0:01.91 htop
369 userftp 20 0 15232 4032 2040 S 0.0 3.3 0:05.62 proftpd: userftp - 192.168.1.4: STOR bigFile.iso
350 root 20 0 10048 3228 2680 S 0.0 2.6 0:00.63 sshd: root@pts/1
1 root 20 0 4724 2700 1860 S 0.0 2.2 0:01.15 /sbin/init console=ttyS0,115200 mtdparts=orion_nand:1M(u-bo
176 root 20 0 8760 1304 924 S 0.0 1.1 0:00.18 /usr/lib/systemd/systemd-udevd
182 root 20 0 7972 2728 2512 S 0.0 2.2 0:00.30 /usr/lib/systemd/systemd-journald
245 root 20 0 4808 1244 736 S 0.0 1.0 0:00.05 /usr/sbin/crond -n
249 root 20 0 6276 2232 1872 S 0.0 1.8 0:00.05 /usr/sbin/sshd -D
250 root 20 0 3104 1284 1100 S 0.0 1.0 0:00.02 /usr/lib/systemd/systemd-logind
251 dbus 20 0 2664 1292 1100 S 0.0 1.1 0:00.04 /usr/bin/dbus-daemon --system --address=systemd: --nofork -
252 root 20 0 7040 2932 2340 S 0.0 2.4 0:00.24 /usr/sbin/syslog-ng -F
266 root 20 0 3784 688 612 S 0.0 0.6 0:00.00 /sbin/agetty --noclear tty1 38400 linux
267 root 20 0 3784 724 648 S 0.0 0.6 0:00.00 /sbin/agetty -s ttyS0 115200 38400 9600 vt102
315 root 20 0 1920 488 380 S 0.0 0.4 0:00.00 dhcpcd -qL -t 10 eth0
322 ntp 20 0 3484 984 752 S 0.0 0.8 0:00.13 /usr/sbin/ntpd -s
323 root 20 0 3652 684 464 S 0.0 0.6 0:00.01 /usr/sbin/ntpd -s
330 root 20 0 10048 3228 2680 S 0.0 2.6 0:00.43 sshd: root@pts/0
332 root 20 0 5340 1916 1536 S 0.0 1.6 0:00.08 -bash
352 root 20 0 5340 1892 1516 S 0.0 1.5 0:00.10 -bash
365 nobody 20 0 13732 1944 876 S 0.0 1.6 0:00.01 proftpd: (accepting connections)
367 userftp 20 0 14836 3464 2028 S 0.0 2.8 0:00.28 proftpd: userftp - 192.168.1.4: IDLE

F1Help F2Setup F3SearchF4FilterF5Tree F6SortByF7Nice -F8Nice +F9Kill F10Quit
')


smb.conf:
$this->bbcode_second_pass_code('', '
[global]
workgroup = WORKGROUP
server string = ALARM
netbios name = ALARM
security = user
encrypt passwords = yes
null passwords = yes
map to guest = bad user
load printers = no
printing = bsd
printcap name = /dev/null
disable spools = yes
log file = /var/log/samba/%m.log
max log size = 50
dns proxy = no
socket options = TCP_NODELAY SO_KEEPALIVE SO_RCVBUF=17520 SO_SNDBUF=17520 IPTOS_LOWDELAY
client ntlmv2 auth = yes


[ArchMediaShare]
path = /home/filesys1
read only = no
public = yes
writable = yes
valid users = root
guest ok = no
delete readonly = yes
oplocks = yes


* - nofile 16384

')

ifconfig output:
$this->bbcode_second_pass_code('', '
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
inet 192.168.1.16 netmask 255.255.255.0 broadcast 192.168.1.255
inet6 fe80::50:43ff:fe7f:110b prefixlen 64 scopeid 0x20<link>
ether 02:50:43:7f:11:0b txqueuelen 1000 (Ethernet)
RX packets 108008 bytes 156081364 (148.8 MiB)
RX errors 0 dropped 97 overruns 0 frame 0
TX packets 88081 bytes 6500656 (6.1 MiB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
device interrupt 11

lo: flags=73<UP,LOOPBACK,RUNNING> mtu 16436
inet 127.0.0.1 netmask 255.0.0.0
inet6 ::1 prefixlen 128 scopeid 0x10<host>
loop txqueuelen 0 (Local Loopback)
RX packets 4 bytes 260 (260.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 4 bytes 260 (260.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

')

parted listing:
$this->bbcode_second_pass_code('', '
GNU Parted 3.1
Using /dev/sda
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) print
Model: ATA ST3000DM001-1CH1 (scsi)
Disk /dev/sda: 3001GB
Sector size (logical/physical): 512B/4096B
Partition Table: gpt
Disk Flags: pmbr_boot

Number Start End Size File system Name Flags
1 32.3kB 20.0GB 20.0GB ext3 Linux filesystem
2 20.0GB 21.1GB 1074MB linux-swap(v1) Linux swap
3 21.1GB 3001GB 2980GB ext4 Linux filesystem

')
Ophidian
 
Posts: 5
Joined: Mon Jan 14, 2013 12:55 am

Re: Terrible intranet performance

Postby Ophidian » Mon Jan 14, 2013 2:22 am

Back to 27 KiB/sec. Sigh.
Ophidian
 
Posts: 5
Joined: Mon Jan 14, 2013 12:55 am

Re: Terrible intranet performance

Postby pepedog » Mon Jan 14, 2013 9:40 am

Bad cable?
pepedog
Developer
 
Posts: 2431
Joined: Mon Jun 07, 2010 3:30 pm
Location: London UK

Re: Terrible intranet performance

Postby Ophidian » Mon Jan 14, 2013 11:01 pm

If I have some time tonight I might try switching the cable and/or reinstalling ALARM. I wonder if I installed incorrectly?
Ophidian
 
Posts: 5
Joined: Mon Jan 14, 2013 12:55 am

Re: Terrible intranet performance

Postby Ophidian » Fri Jan 18, 2013 12:52 am

A quick update for any who are interested:

I've restored the stock bootloader in flash, wondering if my Arch Linux is messing things up. Still slow transfer.

Next, I alter some settings in the Win8 box, disable flow control and set jumbo frames to 4K. I set MTU to 4000 in the GFH. Now I'm getting 22 MB/s, with 91% CPU use in the GFH! Reading back, I get 27MB/s.

I did see higher before I started goofing with the GFH, but this is clearly a step in the right direction. The media partition has been reformatted to ext3, as the stock OS doesn't like ext4. Using the mke2fs and a swap partition, I can format large no problem.

I still don't know what caused the large drop in net performance, though.
Ophidian
 
Posts: 5
Joined: Mon Jan 14, 2013 12:55 am

Re: Terrible intranet performance

Postby moonman » Fri Jan 18, 2013 1:18 am

flow control was the only setting you needed to change. Change mtu and jumbo back to stock as they probably reduce theoughput in this case. Everything else seems right.
Pogoplug V4 | GoFlex Home | Raspberry Pi 4 4GB | CuBox-i4 Pro | ClearFog | BeagleBone Black | Odroid U2 | Odroid C1 | Odroid XU4
-----------------------------------------------------------------------------------------------------------------------
[armv5] Updated U-Boot | [armv5] NAND Rescue System
moonman
Developer
 
Posts: 3388
Joined: Sat Jan 15, 2011 3:36 am


Return to Marvell Kirkwood

Who is online

Users browsing this forum: No registered users and 16 guests