You are not logged in.

#1 2019-01-03 13:05:55

jfabernathy
Member
From: North Carolina
Registered: 2019-01-03
Posts: 123

[Solved]systemd-networkd.service failed with core-dump

This is my first install.  I managed to get to the point where I can boot but when I setup the network I'm stumped.

I looked at the wiki on network configuration. ip link shows dev eno1 is down as expected so I moved on to network manager. I used the systemd one because I think that is the standard one now. So I started systemd-networkd and that failed because a fatal signal was delivered causing the control process to dump core.

systemctl status just confirmed the signal=ABRT

journalctl shows that the service start request repeated too quickly and then the abort.  Sorry I can't share that information but with the network down I can't ssh in and capture the logs.

This install was per the wiki as of yesterday and the kernel is 4.20.0-arch1-ARCH #1 SMP PREEMPT Mon Dec 24 03:00:40 UTC 2018 x86_64

Jim A

Last edited by jfabernathy (2019-01-03 19:32:22)

Offline

#2 2019-01-03 13:52:09

seth
Member
From: Won't reply 2 private help req
Registered: 2012-09-03
Posts: 78,688

Re: [Solved]systemd-networkd.service failed with core-dump

"I think that is the standard one now"
Hardly.

eno1 sounds like a wired connection and if that's all you use, just use dhcpcd - more than sufficient for the task.
https://wiki.archlinux.org/index.php/Dhcpcd
You can just "dhcpcd eno1" for the beginning.
Ensure no other network managing services are running.

If you want to know why systemd-networkd aborted, you can post the coredumps and system statūs and whatnot once you've an established network. Or you do the usb walk.

Offline

#3 2019-01-03 14:44:50

jfabernathy
Member
From: North Carolina
Registered: 2019-01-03
Posts: 123

Re: [Solved]systemd-networkd.service failed with core-dump

At your suggestion I disabled all network managing services and used dhcpcd eno1 to get going. I enabled dhcpcd.service, then installed openssh and enabled it's service. Couple of issues though:
1. after a reboot it takes minutes after the login prompt on the main console before ssh is available to login from a remote compute.
2. There are applications I work with that have used the systemd services to control that they don't start running until after the network is up and they use systemd-networkd-wait-online.service. I'm concerned that I may have a problem with that if i can't get systemd-networkd working.

Offline

#4 2019-01-03 14:49:11

schard
Forum Moderator
From: Hannover
Registered: 2016-05-06
Posts: 2,720
Website

Re: [Solved]systemd-networkd.service failed with core-dump

1) You probably need a random number generator like rngd or haveged.
2) Afaik systemd-networkd-wait-online.service does not require systemd-networkd.

Last edited by schard (2019-01-03 14:50:17)


Inofficial first vice president of the Rust Evangelism Strike Force

Offline

#5 2019-01-03 14:52:28

seth
Member
From: Won't reply 2 private help req
Registered: 2012-09-03
Posts: 78,688

Re: [Solved]systemd-networkd.service failed with core-dump

Or on newer kernels, just pass "random.trust_cpu=on" to the kernel commandline (read up eg. https://www.phoronix.com/scan.php?page= … -RNG-Trust on details)

Offline

#6 2019-01-03 15:14:49

jfabernathy
Member
From: North Carolina
Registered: 2019-01-03
Posts: 123

Re: [Solved]systemd-networkd.service failed with core-dump

The random.trust_cpu=on kernel commandline had no impact.

Offline

#7 2019-01-03 15:17:09

jfabernathy
Member
From: North Carolina
Registered: 2019-01-03
Posts: 123

Re: [Solved]systemd-networkd.service failed with core-dump

However if I login to the console first then ssh login is immediately accepted

Offline

#8 2019-01-03 15:25:21

seth
Member
From: Won't reply 2 private help req
Registered: 2012-09-03
Posts: 78,688

Re: [Solved]systemd-networkd.service failed with core-dump

Sounds a lot like an entropy issue, try installing & enabling haveged and post "cat /proc/cmdline" to check the kernel parameters.

Offline

#9 2019-01-03 15:30:43

jfabernathy
Member
From: North Carolina
Registered: 2019-01-03
Posts: 123

Re: [Solved]systemd-networkd.service failed with core-dump

seth wrote:

Sounds a lot like an entropy issue, try installing & enabling haveged and post "cat /proc/cmdline" to check the kernel parameters.

installing & enabling haveged fixed it.

initrd=\intel-ucode.img initrd=\initramfs-linux.img root=PARTUUID="69333ee4-6342-41c3-8758-5c744c2f72ba" rw random.trust_cpu=on

Offline

#10 2019-01-03 15:39:54

jfabernathy
Member
From: North Carolina
Registered: 2019-01-03
Posts: 123

Re: [Solved]systemd-networkd.service failed with core-dump

should I remove the kernel cmdline "random.trust_cpu=on"

Offline

#11 2019-01-03 16:03:28

seth
Member
From: Won't reply 2 private help req
Registered: 2012-09-03
Posts: 78,688

Re: [Solved]systemd-networkd.service failed with core-dump

I'd rather worry why it doesn't provide the expected entropy.
But since haveged provides it, you can remove the parameter, yes.

Offline

#12 2019-01-03 16:04:23

loqs
Member
Registered: 2014-03-06
Posts: 19,106

Re: [Solved]systemd-networkd.service failed with core-dump

random.trust_cpu=on requires the CPU supports RDSEED or RDRAND (RDSEED is much newer than RDRAND so if the CPU does not support RDRAND it will not support RDSEED)

lscpu | grep --color rdrand

Edit:
@seth could be Sandy Bridge or earlier Intel not sure on AMD.

Last edited by loqs (2019-01-03 16:08:10)

Offline

#13 2019-01-03 16:12:38

jfabernathy
Member
From: North Carolina
Registered: 2019-01-03
Posts: 123

Re: [Solved]systemd-networkd.service failed with core-dump

random.trust_cpu=on is now gone.  ssh still works right after boot and before console is logged into.

lscpu | grep --color rdrand produces nothing.

Thanks,

Offline

Board footer

Powered by FluxBB