You are not logged in.
I'm running into a problem which I'm not able to debug. First, here is my setup:
Motorola Arch Linux
{Internet} --- Surfboard ------------ eno1 --- Server --- enp2so ---------- {Local Net}
Modem
The server is running iptables and NATing for the local network. Most of the time this works great, but intermittently the eno1 interface goes down:
[root@toad eno1]# ip addr show dev eno1
3: eno1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc fq_codel state DOWN group default qlen 1000
link/ether 74:d4:35:18:11:36 brd ff:ff:ff:ff:ff:ff
inet 76.198.131.124/24 brd 67.198.113.255 scope global eno1
valid_lft forever preferred_lft forever
The IP address is static and provided by the ISP:
[root@toad eno1]# cd /etc/netctl
[root@toad netctl]# cat eno1
Description='A basic static ethernet connection'
Interface=eno1
Connection=ethernet
IP=static
Address=('76.198.131.124/24')
#Routes=('192.168.0.0/24 via 192.168.1.2')
Gateway='76.198.131.254'
DNS=('76.198.131.124')
Taking the interface down and bringing it back up does not solve the problem:
# ip link set eno1 down
# ip link set eno1 up
# ip route add default via 76.198.131.254
When the interface is not "crashed" that sequence of events does bring the interface back up in a functional state. When it's crashed, the only thing which brings it back is rebooting the machine.
This makes me suspect a driver problem, but that particular ethernet port is driven by an intel chip, and the intel driver is normally rock solid:
[root@toad eno1]# pwd
/sys/class/net/eno1
[root@toad eno1]# readlink ./device/driver/module
../../../../module/e1000e
It's generally been so intermittent (once every couple of weeks?) that I haven't worried about it much, but lately it's been happening more often and takes down the home network.
Oh, and for the most recent crash the Motorola Surfboard logs show nothing. I recently had everything rewired from the cable access point to the termination in the house, so I know the wires are good as well. The other possibility is an intermittently bad cable connection? But that doesn't explain why a reboot is needed to bring the interface back up.
Offline
Next time this happens I need to also test unloading/loading the e1000e module in addition to taking the interface down. If I can't find a solution, and it works to reload the driver I'll set up a cron job to periodically unload/reload the driver. I'd prefer not to do this, as this will mess with all of my terminals in/out/through the machine.
Last edited by pgoetz (2015-09-05 09:04:36)
Offline