RTnet packet loss issue

Johannes Holtz johannes.holtz at compador.de
Thu Mar 28 09:53:06 CET 2019


Preface:

Xenomai 2.6

Vanilla kernel 3.8.13 i686

RTnet 0.9.13


I have two PCs (M and S) connected via Ethernet exclusively for RTnet use.

M--[switch]---S

M's NIC is driven by the rt_r8169 driver

S's NIC by the rt_e1000 driver


The problem is that the connection is unstable and about 80% of the 
packets get lost.

root at machinectrl:~# rtping 192.168.0.1
Real-time PING 192.168.0.1 56(84) bytes of data.
64 bytes from 192.168.0.1: icmp_seq=1 time=111.2 us
64 bytes from 192.168.0.1: icmp_seq=7 time=99.2 us
64 bytes from 192.168.0.1: icmp_seq=13 time=99.1 us
64 bytes from 192.168.0.1: icmp_seq=23 time=100.5 us
64 bytes from 192.168.0.1: icmp_seq=24 time=97.5 us
64 bytes from 192.168.0.1: icmp_seq=30 time=104.8 us
64 bytes from 192.168.0.1: icmp_seq=32 time=101.8 us
64 bytes from 192.168.0.1: icmp_seq=34 time=102.1 us
64 bytes from 192.168.0.1: icmp_seq=35 time=101.4 us
64 bytes from 192.168.0.1: icmp_seq=37 time=101.5 us
64 bytes from 192.168.0.1: icmp_seq=38 time=99.7 us
64 bytes from 192.168.0.1: icmp_seq=43 time=99.2 us
64 bytes from 192.168.0.1: icmp_seq=54 time=101.3 us
64 bytes from 192.168.0.1: icmp_seq=60 time=102.2 us
64 bytes from 192.168.0.1: icmp_seq=63 time=97.3 us
64 bytes from 192.168.0.1: icmp_seq=69 time=100.3 us
64 bytes from 192.168.0.1: icmp_seq=80 time=101.2 us
64 bytes from 192.168.0.1: icmp_seq=84 time=100.7 us
64 bytes from 192.168.0.1: icmp_seq=86 time=103.4 us
64 bytes from 192.168.0.1: icmp_seq=88 time=105.1 us
64 bytes from 192.168.0.1: icmp_seq=89 time=101.2 us
64 bytes from 192.168.0.1: icmp_seq=96 time=100.7 us
64 bytes from 192.168.0.1: icmp_seq=98 time=99.2 us
64 bytes from 192.168.0.1: icmp_seq=99 time=103.2 us
64 bytes from 192.168.0.1: icmp_seq=113 time=98.8 us
64 bytes from 192.168.0.1: icmp_seq=116 time=105.7 us
64 bytes from 192.168.0.1: icmp_seq=119 time=111.2 us
64 bytes from 192.168.0.1: icmp_seq=126 time=101.7 us
64 bytes from 192.168.0.1: icmp_seq=130 time=102.6 us
^C
--- 192.168.0.1 rtping statistics ---
135 packets transmitted, 29 received, 79% packet loss
worst case rtt = 111.2 us

This happens sometimes from the start after both machines are booted 
without prior communication.
If it is not broken from the start the RTnet works as expected.
But it also happens after the system is running for several hours 
without any noticeable user action that would cause it.

The cables and connectors have all been double checked.

Is this a know error or is there anything you can tell me about this 
pattern?

I'm grateful for every piece of advice as always.


Cheers,

Johannes





More information about the Xenomai mailing list