LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Networking (https://www.linuxquestions.org/questions/linux-networking-3/)
-   -   awfull network speeds (https://www.linuxquestions.org/questions/linux-networking-3/awfull-network-speeds-918556/)

JonathanWilson 12-13-2011 02:10 PM

awfull network speeds
 
I have 5 machines on my network (6 including router), 4 of which are 1gig and 2 are 10/100 (a laptop and the router)

My 8 port network is 1gig and shows the conections with a multi colour led. It shows the 1gig conections are 1gig.

The following is reported on my file server (pc 1gig) and my desktop (pc 1gig)

One runs debian squeeze and the other ubuntu 10.10

The hardware is asus p43 and asus p45 respectively and both have realtek rtl8168d on board Ethernet.


Quote:

ethtool eth0
Settings for eth0:
Supported ports: [ TP MII ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Half 1000baseT/Full
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Half 1000baseT/Full
Advertised pause frame use: No
Advertised auto-negotiation: Yes
Link partner advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Half 1000baseT/Full
Link partner advertised pause frame use: No
Link partner advertised auto-negotiation: Yes
Speed: 1000Mb/s
Duplex: Full
Port: MII
PHYAD: 0
Transceiver: internal
Auto-negotiation: on
Supports Wake-on: pumbg
Wake-on: g
Current message level: 0x00000033 (51)
Link detected: yes

ifconfig
eth0 Link encap:Ethernet HWaddr 00:19:66:b4:73:2e
inet addr:192.168.1.69 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::219:66ff:feb4:732e/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:2242361 errors:0 dropped:0 overruns:0 frame:0
TX packets:932331 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3075266551 (2.8 GiB) TX bytes:401921089 (383.3 MiB)
Interrupt:28 Base address:0x4000

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:198 errors:0 dropped:0 overruns:0 frame:0
TX packets:198 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:14360 (14.0 KiB) TX bytes:14360 (14.0 KiB)



Quote:

ethtool eth0
Settings for eth0:
Supported ports: [ TP ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Advertised pause frame use: Symmetric Receive-only
Advertised auto-negotiation: Yes
Speed: 1000Mb/s
Duplex: Full
Port: Twisted Pair
PHYAD: 0
Transceiver: internal
Auto-negotiation: on
MDI-X: Unknown
Supports Wake-on: pumbg
Wake-on: g
Current message level: 0x00000033 (51)
Link detected: yes


ifconfig
eth0 Link encap:Ethernet HWaddr 00:19:66:b9:d9:0e
inet addr:192.168.1.66 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::219:66ff:feb9:d90e/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1186742 errors:0 dropped:0 overruns:0 frame:0
TX packets:2446925 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:709108452 (709.1 MB) TX bytes:3131906546 (3.1 GB)
Interrupt:43

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:1511815 errors:0 dropped:0 overruns:0 frame:0
TX packets:1511815 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:17140929424 (17.1 GB) TX bytes:17140929424 (17.1 GB)


The following is reported by netperf

Quote:

root@borgCUBE:/mnt/md6mnt# netperf -D -H 192.168.1.66 -f M
TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.1.66 (192.168.1.66) port 0 AF_INET : demo
Interim result: 3.42 MBytes/s over 4.50 seconds
Interim result: 2.18 MBytes/s over 1.57 seconds
Interim result: 3.29 MBytes/s over 2.62 seconds
Interim result: 3.12 MBytes/s over 1.14 seconds
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. MBytes/sec

87380 16384 16384 10.31 3.12
Quote:

root@AOC:~# netperf -D -H 192.168.1.69 -f M
TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.1.69 (192.168.1.69) port 0 AF_INET : demo
Interim result: 4.58 MBytes/s over 2.50 seconds
Interim result: 4.32 MBytes/s over 1.24 seconds
Interim result: 3.43 MBytes/s over 1.26 seconds
Interim result: 3.38 MBytes/s over 1.01 seconds
Interim result: 1.57 MBytes/s over 2.15 seconds
Interim result: 2.36 MBytes/s over 1.22 seconds
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. MBytes/sec

87380 16384 16384 10.00 3.25
Which seems total wrong as surely it should surely be a factor faster than reported.

Any help much appreciated in finding a solution

T3RM1NVT0R 12-13-2011 02:40 PM

@ Reply
 
Hi JonathanWilson,

It would be great if you would have given the IP address information for these machines and would have let us know the communication flow that occurs between these machines.

The output that you pasted shows that the system's are configured to use "Auto Negotiation" which according to me is the reason behind the slowness. If you could hard code them to use 1 gig speed then it will work fine and when they are communicating with 10/100 base machine the speed will automatically come down to match their speed.

It appears to me that as of now the are communicating using 10/100 base speed by default.

JonathanWilson 12-14-2011 12:53 PM

I am begining to think its a network card problem in my desktop as the speeds are slow for my desktop to/from server and fairly fast from other machines to/from on the network.

setting on my desktop to:

ethtool -s eth0 speed 10 duplex full autoneg off

results in iperf results of :

Quote:

root@borgCUBE:~# iperf -c 192.168.1.66 -f M -i 2
------------------------------------------------------------
Client connecting to 192.168.1.66, TCP port 5001
TCP window size: 0.02 MByte (default)
------------------------------------------------------------
[ 3] local 192.168.1.69 port 44378 connected with 192.168.1.66 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 0.70 MBytes 0.35 MBytes/sec
[ 3] 2.0- 4.0 sec 1.01 MBytes 0.50 MBytes/sec
[ 3] 4.0- 6.0 sec 0.87 MBytes 0.43 MBytes/sec
[ 3] 6.0- 8.0 sec 1.02 MBytes 0.51 MBytes/sec
[ 3] 8.0-10.0 sec 0.75 MBytes 0.38 MBytes/sec
[ 3] 0.0-10.0 sec 4.34 MBytes 0.43 MBytes/sec
setting to 100 results in :

Quote:

root@borgCUBE:~# iperf -c 192.168.1.66 -f M -i 2
------------------------------------------------------------
Client connecting to 192.168.1.66, TCP port 5001
TCP window size: 0.02 MByte (default)
------------------------------------------------------------
[ 3] local 192.168.1.69 port 44387 connected with 192.168.1.66 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 1.80 MBytes 0.90 MBytes/sec
[ 3] 2.0- 4.0 sec 3.50 MBytes 1.75 MBytes/sec
[ 3] 4.0- 6.0 sec 4.01 MBytes 2.00 MBytes/sec
[ 3] 6.0- 8.0 sec 1.68 MBytes 0.84 MBytes/sec
[ 3] 8.0-10.0 sec 3.71 MBytes 1.86 MBytes/sec
[ 3] 0.0-10.0 sec 14.7 MBytes 1.46 MBytes/sec

If i try to set autoneg off (on my desktop) when setting speed 1000 it tells me invalid argument.

Quote:

root@AOC:~# ethtool -s eth0 speed 1000 duplex full autoneg off
Cannot set new settings: Invalid argument
not setting speed
not setting duplex
not setting autoneg
Quote:

ethtool -s eth0 speed 1000 duplex full autoneg on
I did check to see if the same error happened on my server and it did, so autoneg cannot be of for 1000 speeds.

As a quick and dirty test I went to to my windows laptop (10/100) and sent a 1 gig file to both the server and my desktop and the results were

10.4 MB/s server
5.82 MB/s desktop

Cedrik 12-14-2011 01:56 PM

Does speed change if you disconnect the laptop from the 8 port switch ?

T3RM1NVT0R 12-14-2011 02:13 PM

@ Reply
 
What is the speed of your server's network adapter. If it is a gigbit adapter then yes there is a possibility of network card issue on your desktop. Because from the output you pasted your Desktop has got a gigabit adapter and your laptop has got base 100 full duplex adapter. So if laptop is able to transfer files quickly to your server and not to your desktop then there can be an issue with desktop's NIC.

Could you please let us know what are the OSs installed on these three machines (server, laptop and your desktop)

JonathanWilson 12-15-2011 07:53 AM

@Cedrik

I will try removing everything that is 10/100 (laptop and ISP router) and see what speeds I get from my lads pc (gigabit, windows) although the 8 port hub should be able to handle mixed speeds.


@t3em1nvt0r

The server is debian setup with ssh and samba and headless, and yes its a gigabit builtin adapter
The desktop is Ubuntu 10.10
And the laptop is Windows vista.

I think a couple of other things I will check are:
swap network cable from the laptop to the desktop to test the cable,
boot from a clean install of debian on the desktop, just to be sure its not software related as the ubuntu has had numerious updates and things installed and removed over time.

JonathanWilson 12-16-2011 11:44 AM

Well that was interesting, the problem was with the cables.

Using a brand new cat 6 cable, I direct connected my pc and the borgCUBE and my results were:

Quote:

root@borgCUBE:~# iperf -c 192.168.1.69 -f M -i 2
------------------------------------------------------------
Client connecting to 192.168.1.69, TCP port 5001
TCP window size: 0.02 MByte (default)
------------------------------------------------------------
[ 3] local 192.168.1.25 port 46953 connected with 192.168.1.69 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 212 MBytes 106 MBytes/sec
[ 3] 2.0- 4.0 sec 220 MBytes 110 MBytes/sec
[ 3] 4.0- 6.0 sec 222 MBytes 111 MBytes/sec
[ 3] 6.0- 8.0 sec 223 MBytes 111 MBytes/sec
[ 3] 8.0-10.0 sec 222 MBytes 111 MBytes/sec
[ 3] 0.0-10.0 sec 1098 MBytes 110 MBytes/sec
It was not initialy easy to do on the server, however the use of

ifconfig eth0 192.168.1.25 netmask 255.255.255.0 broadcast 192.168.1.255

on the server seemed to keep even after disconnecting the cable, and inserting the direct connection.

the same command (IP changed) on the desktop, and bingo I could ping and then start the iperf test.

I guess its time to go and get some new cat 6 cables for the rest of the computers, then worst case a new gig hub, best case it all goes well and is fast :-)


All times are GMT -5. The time now is 02:38 AM.