sar -n EDEV shows rxfram errors

[Geert Van Pamel]

I am running a Fedora server. I see "Ethernet Receive Framing errors" on my server. Still my server seems to respond normally?

sar -n EDEV
Linux 2.6.11-1.1369_FC4 01/09/2008

12:00:01 AM IFACE rxerr/s txerr/s coll/s rxdrop/s txdrop/s txcarr/s rxfram/s rxfifo/s txfifo/s
12:10:02 AM eth0 0.00 0.00 0.00 0.00 0.00 0.00 4881.82 0.00 0.00
Average: eth0 0.00 0.00 0.00 0.00 0.00 0.00 4885.26 0.00 0.00

Questions:

1. What could be the reason (server? router? other machines on the LAN?)
2. Does it harm (network) performance?
3. Can I do something about it?
4. Would I need to change ethernet NIC settings?
5. Any further troubleshooting tools?

[Vitaly Karasik_1]

- Can we see your network statistic? For example, do you have huge traffic?
- it is may be OT, but I'll suggest to upgrade your server to CENTOS5.

[Ralph Grothe]

Does your NIC driver support stats dumps like
e.g.
# ethtool -S eth0

[Geert Van Pamel]

Not much network traffic!

sar -n DEV |grep eth0
10:00:03 PM IFACE rxpck/s txpck/s rxbyt/s txbyt/s rxcmp/s txcmp/s rxmcst/s
10:10:01 PM eth0 3.25 0.59 268.22 351.46 0.00 0.00 0.00
Average: eth0 3.31 0.53 298.50 268.83 0.00 0.00 0.00

sar -n EDEV |grep eth0

10:20:02 PM eth0 0.00 0.00 0.00 0.00 0.00 0.00 4879.05 0.00 0.00
Average: eth0 0.00 0.00 0.00 0.00 0.00 0.00 4885.19 0.00 0.00


ethtool -S eth0
NIC statistics:
rx_packets: 5173795
tx_packets: 979215
rx_bytes: 537243623
tx_bytes: 573661363
rx_errors: 0
tx_errors: 0
rx_dropped: 9769
tx_dropped: 0
multicast: 0
collisions: 0
rx_length_errors: 1353266787
rx_over_errors: 0
rx_crc_errors: 0
rx_frame_errors: 0
rx_fifo_errors: 0
rx_missed_errors: 0
tx_aborted_errors: 0
tx_carrier_errors: 0
tx_fifo_errors: 0
tx_heartbeat_errors: 0
tx_window_errors: 0
tx_deferred: 0
tx_single_collisions: 0
tx_multi_collisions: 0
tx_flow_control_pause: 0
rx_flow_control_pause: 0
rx_flow_control_unsupported: 0
tx_tco_packets: 0
rx_tco_packets: 0

ifconfig eth0
eth0 Link encap:Ethernet HWaddr 00:90:27:BC:51:2C
inet addr:xxx.xxx.xxx.xxx Bcast:xxx.xxx.xxx.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1464 Metric:1
RX packets:5173986 errors:0 dropped:9769 overruns:0 frame:1353481705
TX packets:979256 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:537260142 (512.3 MiB) TX bytes:573674109 (547.0 MiB)

The MTU was not originally set to the standard ethernet value of 1500 byte.

I will try to set the MTU to the standard value, and monitor if the errors still increase.

ifconfig eth0 mtu 1500

vi /etc/sysconfig/network-scripts/ifcfg-eth0
...
#MTU=1464


But the errors still increase !!

ethtool -S eth0 |grep rx_length_errors
rx_length_errors: 1355337815

ethtool -S eth0 |grep rx_length_errors
rx_length_errors: 1355367122

I am still running IPv4.

Could the reason be external to the server, e.g. other machines on the LAN generating (very) big IPv6 packets?

So in that case I suspect I cannot do anything at all? And it does not harm?

[VK2COT]

Hello,

a) Could you provide results of command:

ethtool -t eth0 (test the adapter card)
ethtool eth0
lspci -v
ping -s 1450 -f your_gw
ping -s 1500 -f your_gw

Make sure to break the ping(1) command
after 10-20 seconds. It will then summarize
the results.

b) Also, check the port setup on the
router/switch side.

There must not be duplex, speed, or
negotiation mismatch!

c) Is your cable very long?

d) It is worth analysing where these
packets come from. If the sender assumes
it can send large packets, and the receiver
doesn't expect them it might cause it.

Use some packet sniffers (tcpdump,
etherape, argus, ntop, or others).

e) By the way, your Fedora 4,
patches and kernel version is a bit
outdated...

Frame errors are typically caused by
incorrect CRC and a non-integer number of
bytes are received. This is usually the
result of collisions or a bad Ethernet
device.

Length Errors are generated when the received
frame length was less than or exceeded the
Ethernet standard. This is most frequently due to incompatible duplex settings.

Cheers,

VK2COT

[Geert Van Pamel]

When the framing errors occurred, IPv6 was still active.

ifconfig
eth0 Link encap:Ethernet HWaddr 00:90:27:BC:51:2C
inet addr:xxx.xxx.xxx.xxx Bcast:xxx.xxx.xxx.255 Mask:255.255.255.0
inet6 addr: xxxx::xxx:xxxx:xxxx:xxxx/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:64712446 errors:0 dropped:111195 overruns:131 frame:3709872387
TX packets:8401513 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1713061415 (1.5 GiB) TX bytes:1347360583 (1.2 GiB)

I do not need IPv6, so I disable it during system startup:


cat /etc/rc.local
#!/bin/sh

touch /var/lock/subsys/local

case "$1" in
start)
ifconfig eth0 del xxxx::xxx:xxxx:xxxx:xxxx/64
;;
*)
echo "Usage: $0 {start|stop|status|restart|reload}"
exit 1
esac

exit 0


The system rebooted since then...

uptime
10:15:39 up 19 days, 4:49, 1 user, load average: 1.18, 1.22, 1.19

ifconfig
eth0 Link encap:Ethernet HWaddr 00:90:27:BC:51:2C
inet addr:193.190.246.240 Bcast:193.190.246.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:7396686 errors:0 dropped:0 overruns:0 frame:0
TX packets:1156959 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:681546047 (649.9 MiB) TX bytes:730976501 (697.1 MiB)


Conclusion: The framing errors disappeared by disabling the IPv6 address ... IPv6 ethernet packets are bigger than 1500 byte and appear to trigger fictitious framing errors !