potential bug: buster: systemd-networkd not working reliable
Hi,
it seems systemd-networkd of the Buster distribution does not work
reliable on some of my systems.
Can you help me to further analyze the problem in front of bug creation?
Short summary:
If an interface uses static addresses and the cable is unplugged and
plugged in again, the interface does not go up anymore.
Detailed summary:
Architecture: amd64
Kernel: Linux 4.19.0-6-amd64 #1 SMP Debian 4.19.67-2 (2019-08-28)
x86_64 GNU/Linux
systemd: Version: 241-7~deb10u1
The system contains two ethernet links "enp1s0" and "eno1".
The kernel module used for this interfaces is "igb".
===
Oct 08 09:38:44 sb-0001c0237244 kernel: igb: Intel(R) Gigabit Ethernet
Network Driver - version 5.4.0-k
Oct 08 09:38:44 sb-0001c0237244 kernel: igb: Copyright (c) 2007-2014
Intel Corporation.
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:01:00.0 enp1s0:
renamed from eth0
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:01:00.0: added PHC on enp1s0
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:01:00.0: Intel(R)
Gigabit Ethernet Network Connection
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:01:00.0: enp1s0:
(PCIe:2.5Gb/s:Width x1) 00:01:c0:23:72:44
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:01:00.0: enp1s0: PBA
No: FFFFFF-0FF
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:01:00.0: Using MSI-X
interrupts. 2 rx queue(s), 2 tx queue(s)
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:02:00.0 eno1: renamed from eth0
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:02:00.0: added PHC on eno1
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:02:00.0: Intel(R)
Gigabit Ethernet Network Connection
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:02:00.0: eno1:
(PCIe:2.5Gb/s:Width x1) 00:01:c0:23:72:51
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:02:00.0: eno1: PBA
No: FFFFFF-0FF
Oct 08 09:38:44 sb-0001c0237244 kernel: igb 0000:02:00.0: Using MSI-X
interrupts. 2 rx queue(s), 2 tx queue(s)
Oct 08 09:38:47 sb-0001c0237244 kernel: igb 0000:01:00.0 enp1s0: igb:
enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX/TX
Oct 08 09:38:48 sb-0001c0237244 kernel: igb 0000:02:00.0 eno1: igb:
eno1 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX/TX
===
The network configuration for "eno1" looks like (10-ethernet-eno1.network)
===
[Match]
Name=eno1
[Network]
DHCP=no
Address=10.10.69.1/24
Address=10.10.69.200/24
Address=10.10.69.201/24
Address=10.10.69.202/24
===
The network configuration for "enp1s0" looks like (10-ethernet-enp1s0.network)
===
[Match]
Name=enp1s0
[Network]
DHCP=ipv4
===
If the network cable on enp1s0 is unplugged the assigned address is
dropped and the interface went down. If the cable is plugged in again,
the interface goes up and requests an address using DHCP.
If the network cable on eno1 is unplugged the assigned addresses are
dropped and the interface went down. If the cable is plugged in again,
the interface does NOT go up but stays down.
I enabled the debug log level for systemd-networkd.
If the cable of eno1 is unplugged I see:
===
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Flags
change: -LOWER_UP -RUNNING
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: Sent message
type=signal sender=n/a destination=n/a
path=/org/freedesktop/network1/link/_35
interface=org.freedesktop.DBus.Properties member=PropertiesChanged
cookie=37 reply_cookie=0 signature=sa{sv}as error-name=n/a
error-message=n/a
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Lost carrier
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: NDISC: Stopping
IPv6 Router Solicitation client
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address 10.10.69.202
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address 10.10.69.200
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address 10.10.69.201
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address 10.10.69.1
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: State is
configured, dropping config
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address: 10.10.69.202/24 (valid forever)
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address: 10.10.69.200/24 (valid forever)
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address: 10.10.69.201/24 (valid forever)
Oct 08 09:45:26 sb-0001c0237244 systemd-networkd[294]: eno1: Removing
address: 10.10.69.1/24 (valid forever)
===
If the cable is plugged in again, no additional message appears.
Before I updated to Buster I run into this bug:
https://github.com/systemd/systemd/issues/5625
This could be worked around using: "IPv6AcceptRA=no" and
"LinkLocalAddressing=no"
I don't run into this error after updating anymore (it has been about DHCP).
Using that lines (regardless if no or yes is choosen) for the static
interface on Buster does not make any difference, so the error is
different.
But while reading that bug, I assume the NDISC message above could be
a signal that something went wrong...
Upstream accepts bug reports only for the two most recents systemd
version (243 and 242), so I cannot report it upstream.
Can you help?
Best regards,
Markus
Reply to: