Bad ping/network performance

Z

Zer0

Hello,

I just installed FreeBSD 9 amd64 on an Asus P8Z77-M Motherboard with the Intel i3-3200 CPU. For networking I use a Intel Pro/1000GT. I also tried an older USRobotics Realtek card, but same thing there.

The problem is, that there are some problems with the network connection, I have pings around 1000 to 3000ms and therefore downloading or ssh sessions are nearly impossible.

Code: 
PING google.de (173.194.69.94): 56 data bytes
64 bytes from 173.194.69.94: icmp_seq=0 ttl=50 time=3998.035 ms
64 bytes from 173.194.69.94: icmp_seq=1 ttl=50 time=2998.279 ms
64 bytes from 173.194.69.94: icmp_seq=2 ttl=50 time=1997.430 ms
64 bytes from 173.194.69.94: icmp_seq=3 ttl=50 time=996.550 ms
64 bytes from 173.194.69.94: icmp_seq=4 ttl=50 time=3992.210 ms
64 bytes from 173.194.69.94: icmp_seq=5 ttl=50 time=2992.898 ms
64 bytes from 173.194.69.94: icmp_seq=6 ttl=50 time=1992.039 ms
64 bytes from 173.194.69.94: icmp_seq=7 ttl=50 time=991.127 ms
64 bytes from 173.194.69.94: icmp_seq=8 ttl=50 time=3988.262 ms
64 bytes from 173.194.69.94: icmp_seq=9 ttl=50 time=2989.221 ms
64 bytes from 173.194.69.94: icmp_seq=10 ttl=50 time=1988.325 ms

PING 192.168.178.1 (192.168.178.1): 56 data bytes
64 bytes from 192.168.178.1: icmp_seq=0 ttl=64 time=3988.928 ms
64 bytes from 192.168.178.1: icmp_seq=1 ttl=64 time=2990.695 ms
64 bytes from 192.168.178.1: icmp_seq=2 ttl=64 time=1989.759 ms
64 bytes from 192.168.178.1: icmp_seq=3 ttl=64 time=988.842 ms
64 bytes from 192.168.178.1: icmp_seq=4 ttl=64 time=3985.290 ms
64 bytes from 192.168.178.1: icmp_seq=5 ttl=64 time=2986.033 ms
64 bytes from 192.168.178.1: icmp_seq=6 ttl=64 time=1985.200 ms
64 bytes from 192.168.178.1: icmp_seq=7 ttl=64 time=984.180 ms
64 bytes from 192.168.178.1: icmp_seq=8 ttl=64 time=3981.302 ms
64 bytes from 192.168.178.1: icmp_seq=9 ttl=64 time=2982.069 ms
64 bytes from 192.168.178.1: icmp_seq=10 ttl=64 time=1981.054 ms

localhost has no excessive ping responses:
PING localhost (127.0.0.1): 56 data bytes
64 bytes from 127.0.0.1: icmp_seq=0 ttl=64 time=0.201 ms
64 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=0.116 ms
64 bytes from 127.0.0.1: icmp_seq=2 ttl=64 time=0.125 ms
64 bytes from 127.0.0.1: icmp_seq=3 ttl=64 time=0.113 ms
64 bytes from 127.0.0.1: icmp_seq=4 ttl=64 time=0.090 ms
64 bytes from 127.0.0.1: icmp_seq=5 ttl=64 time=0.155 ms
64 bytes from 127.0.0.1: icmp_seq=6 ttl=64 time=0.125 ms
64 bytes from 127.0.0.1: icmp_seq=7 ttl=64 time=0.113 ms
64 bytes from 127.0.0.1: icmp_seq=8 ttl=64 time=0.089 ms
64 bytes from 127.0.0.1: icmp_seq=9 ttl=64 time=0.117 ms
64 bytes from 127.0.0.1: icmp_seq=10 ttl=64 time=0.091 ms


My ifconfig:
Code: 
em0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
        options=209b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,WOL_MAGIC>
        ether 90:e2:ba:0c:8e:33
        inet 192.168.178.100 netmask 0xffffff00 broadcast 192.168.178.255
        inet6 fe80::92e2:baff:fe0c:8e33%em0 prefixlen 64 scopeid 0x3
        nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
        media: Ethernet autoselect (1000baseT <full-duplex>)
        status: active
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384
        options=3<RXCSUM,TXCSUM>
        inet6 ::1 prefixlen 128
        inet6 fe80::1%lo0 prefixlen 64 scopeid 0x5
        inet 127.0.0.1 netmask 0xff000000
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>

netstat for em0 shows no errors or drops on the network:
Code: 
Name    Mtu Network       Address              Ipkts Ierrs Idrop    Opkts Oerrs  Coll
usbus     0 <Link#1>                               0     0     0        0     0     0
usbus     0 <Link#2>                               0     0     0        0     0     0
em0    1500 <Link#3>      90:e2:ba:0c:8e:33     4995     0     0     4363     0     0
em0    1500 192.168.178.0 192.168.178.100       4695     -     -     4175     -     -
em0    1500 fe80::92e2:ba fe80::92e2:baff:f        0     -     -        1     -     -
usbus     0 <Link#4>                               0     0     0        0     0     0
lo0   16384 <Link#5>                               0     0     0        0     0     0
lo0   16384 localhost     ::1                      0     -     -        0     -     -
lo0   16384 fe80::1%lo0   fe80::1                  0     -     -        0     -     -
lo0   16384 your-net      localhost                0     -     -        0     -     -

I tested the hardware with another linux operating system and networking runs just fine.

Any help is appreciated :)

Regards, Paul
 
Try a different cable and/or switch port. Also verify if there aren't any errors on the switch.
 
Install net/mtr-nox11, run [cmd=]mtr 173.194.69.94[/cmd] and see where the actual degradation takes place. If it's always within the local network, then (indeed) check cables, switch ports.
 
The first thing I tried, was using a different cable and I plugged the network directly into the router. I will try the mtr tool and get back to you.
 
Looking at the pings you already posted I'm quite sure the issue is with your local network. The pings to 192.168.178.1 (I assume that's your router) are already quite high.
 
I traced the connection over my router:

First to my router
# mtr 192.168.178.1
Code: 
                             My traceroute  [v0.82]
Arcos (0.0.0.0)                                        Wed Apr 18 15:37:49 2012
Keys:  Help   Display mode   Restart statistics   Order of fields   quit
                                       Packets               Pings
 Host                                Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 192.168.178.1                     5.9%    18  727.3 2119. 325.2 3746. 1155.
The corresponding wireshark tracefile http://www.mediafire.com/?2pucd99641698rw

and then to google
# mtr 173.194.69.94
Code: 
                             My traceroute  [v0.82]
Arcos (0.0.0.0)                                        Wed Apr 18 15:35:31 2012
Keys:  Help   Display mode   Restart statistics   Order of fields   quit
                                       Packets               Pings
 Host                                Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 192.168.178.1                    71.4%    43  2276. 1903.  22.4 3337. 1123.
    87.186.224.34
87.187.186.224.34                    73.8%    43  2184. 2323. 1048. 3921. 947.6
87.187.190.167.34                    69.0%    43  2092. 2189.  72.2 3911. 1169.
hh-ehh-ea7-i.HH.DE.NET.DTAG.DE   26.073.8%    43  2000. 2646. 796.9 3996. 992.0
80.180.150.171.130                   71.4%    43  1908. 2661. 671.0 3932. 1032.
209.209.85.249.132                   69.0%    43  1816. 2426. 545.0 3840. 1086.
72.172.14.233.216                    69.0%    43  1724. 2328. 419.1 3748. 1088.
209.209.85.240.154                   73.8%    43  1632. 2540. 1300. 3656. 856.0
64.264.233.174.29                    78.9%    39  1540. 2537. 1540. 3565. 848.8
    64.233.174.53
10. ???
bk-ibk-in-f94.1e100.net               2.6%    39  783.3 1867.  45.0 3819. 1123.
and the wireshark tracefile http://www.mediafire.com/?z0j437wrqvls8d7

In both files, you will see some traffic from pkg_add, but the main thing I notice, is, that the google pings generate lots of ttl exceeded in transit. Other then that, at least to me nothing suspicious.

I don't get ttl exceeded messages from the pings to my router, but the ping is way higher then normal, but why? I don't have these problems with my Windows PC and my debian laptop. Also the network connection runs fine if I use debian on the PC in question here. I hope you have an idea here.
 
Try [cmd=]mtr -r -c 60 somehost[/cmd] for better output. It will give you a nice report after 60 seconds, which will look better.
 
mtr -r -c 60 192.168.178.1
Code: 
HOST: Arcos                       Loss%   Snt   Last   Avg  Best  Wrst StDev
  1.|-- 192.168.178.1              5.0%    60  671.2 2189. 155.1 3929. 1152.2

mtr -r -c 60 193.99.144.80
Code: 
HOST: Arcos                       Loss%   Snt   Last   Avg  Best  Wrst StDev
  1.|-- 192.168.178.1             81.7%    60  3034. 1768.   6.1 4104. 1287.9
    |  `|-- 87.186.224.34
  2.|-- 87.186.224.34             81.7%    60  2956. 2403. 870.6 4026. 1222.3
  3.|-- 87.190.167.46             81.7%    60  2878. 2311. 726.6 3948. 1224.5
  4.|-- f-ed4-i.F.DE.NET.DTAG.DE  81.7%    60  2800. 2219. 582.6 3870. 1227.2
  5.|-- 217.243.218.38            83.3%    60  2722. 2259. 438.7 3792. 1212.0
  6.|-- heise2.f.de.plusline.net  81.7%    60  2644. 2398. 294.7 4728. 1390.1
  7.|-- ???                       100.0    60    0.0   0.0   0.0   0.0   0.0

# vmstat -i
Code: 
interrupt                          total       rate
irq16: em0 xhci0+                  21094          1
irq23: ehci1                       31492          2
cpu0:timer                       1769602        124
irq256: hdac0                         48          0
irq257: ahci0                     590988         41
cpu1:timer                        292682         20
cpu3:timer                        365865         25
cpu2:timer                       1029205         72
Total                            4100976        288

# sysctl net.inet.ip.intr_queue_drops
Code: 
net.inet.ip.intr_queue_drops: 0
 
What's weird about that? AMD64 just means, that I installed the 64bit version of FreeBSD. I can't install ia64, cause that is for Intel Itanium CPUs and why should I install i386 when I want to use 16GB of RAM and don't want to use PAE. Or did I miss something?
 
There is a Realtek onboard, but it doesn't work out of the box (neither with FreeBSD nor Debian) and is currently deactivated. Maybe I can get it working, but I'm not that familiar with FreeBSD as I'm with Debian, so I might need some time. I will try it tomorrow cause I'm at work right now.
 
Also, if you have any, try to ping another machine on your network, not the router. As it may be the router that's causing the delays.
 
Here is a mtr ping from the FreeBSD box to my local Windows machine:
Code: 
HOST: Arcos                       Loss%   Snt   Last   Avg  Best  Wrst StDev
  1.|-- 192.168.178.21             6.7%    60  796.8 2322. 796.8 3847. 1126.5

And here a ping using WinMTR from the Windows PC to the BSD box:
Code: 
|------------------------------------------------------------------------------------------|
|                                      WinMTR statistics                                   |
|                       Host              -   %  | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
|                         192.168.178.100 -    0 |   78 |   78 |    0 | 1690 | 3999 |    0 |
|________________________________________________|______|______|______|______|______|______|

Also ssh is nearly unusable, it's like I would connect over a 4800 baud line or worse.

I'm sure it has nothing to do with my local network. The only common denominator for this problem is the FreeBSD installation. Debian and Windows installations running smooth on the same hardware, without any modification to the hardware or the network. Only FreeBSD has this strange network problems. But since I'm not the FreeBSD or network guru, I hope you guys here have an idea on how to tackle this :)
 
Only a wild guess, however, it's worth to give it a quick try.

Switch, net.inet.tcp.delayed_ack to 0, since the equivalent out-of-the-box setting on Windows is also 0. If it doesn't help, then set it back to 1.
 
It's a new board, slighty different architecture. Just an upgrade from PCI2 to PCI3 bus may raise incompatibility to kernel or device drivers. In fact that internal network card is not working in the first place might be an indication of it.

Come on, FreeBSD is not known as a very welcomer to new hardware.
 
WinMTR from my Windows Box:
Code: 
|------------------------------------------------------------------------------------------|
|                                      WinMTR statistics                                   |
|                       Host              -   %  | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
|                           192.168.178.1 -    0 |  130 |  130 |    0 |    1 |   18 |    1 |
|________________________________________________|______|______|______|______|______|______|
 
em0 shares its IRQ with xhci and other(s) device(s), maybe this is the root cause of your problem, try to use MSI or disable USB and see how it goes.
 
I disabled as much as I could, but USB can't be completely disabled
vmstat -i
Code: 
interrupt                          total       rate
irq16: em0 ehci0                    1051          1
irq23: ehci1                        1715          2
cpu0:timer                         51069         73
irq256: mps0                         371          0
irq257: ahci0                       1073          1
cpu1:timer                          2471          3
cpu3:timer                          3022          4
cpu2:timer                          8827         12
Total                              69599        100
Is there a way to kind of remap the IRQs?
 
You must log in or register to reply here.
Back
Top