Issue booting CoreOS from PXE to bare metal devices

[Denis Fayruzov]

Hi all,

I'm having issues loading "default" CoreOS from example published at: https://coreos.com/os/docs/latest/booting-with-pxe.html 

The trouble is this same setup is working perfectly for booting Hyper-V VMs, but bare metal devices (2 laptops and server) are falling to Emergency Console.

Hardware setup for both scenarios:

1) Router serving as DHCP / TFTP / PXE server -> Laptop Ethernet port ->  Hyper-V VMs on external (bridged) virtual network adapter

2) Router serving as DHCP / TFTP / PXE server -> Ethernet connected bare metal devices  

Ignition config is served by TFTP for both cases:

Label CoreOS

        MENU DEFAULT

        kernel /mnt/sda1/tftp/disks/coreos/coreos_production_pxe.vmlinuz

        initrd /mnt/sda1/tftp/disks/coreos/coreos_production_pxe_image.cpio.gz

        append coreos.first_boot=1 coreos.config.url=tftp://192.168.1.1/mnt/sda1/tftp/disks/coreos/pxe-config.ign

In second scenario I see two errors  in log that can point to the failure reason:

[    7.170688] localhost ignition[250]: failed to fetch config: write udp [::]:53979->192.168.1.1:69: sendto: network is unreachable

[    7.171621] localhost ignition[250]: failed to acquire config: write udp [::]:53979->192.168.1.1:69: sendto: network is unreachable

I attached full rdsosreport.txt log to check if I missed something, I can do the same for first scenario, when VMs are booting fine.

Any thoughts/suggestions will be appreciated greatly.

BR,

Denis

[Benjamin Runnels]

I am seeing the same thing.  Did you ever find a resolution?

[Benjamin Runnels]

Putting same config on pastebin and setting coreos.config.url to the http raw link works.  I verified I can download the file from the url using tftp client.

On Friday, April 20, 2018 at 3:44:01 AM UTC-5, Denis Fayruzov wrote:

[Seán C. McCord]

The "network unreachable" error is pretty specific.  It means that the network on which resides 192.168.1.1 is not reachable from the client.  This could be due to the network not being up, having improper subnet mask set, not having a route (such as a default gateway) to the network, etc.  You can ignore issues with the TFTP service itself: it isn't getting that far.  You can also ignore the specific syntax of the kernel option,  Your problem is in the network and/or its configuration.

The first thing to do would be to add the 'coreos.autologin' option to the kernel parameters, then check that the network configuration is correct.

--
You received this message because you are subscribed to the Google Groups "CoreOS User" group.
To unsubscribe from this group and stop receiving emails from it, send an email to coreos-user...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Seán C McCord

CyCore Systems, Inc

+1 888 240 0308

PGP/GPG: http://cycoresys.com/scm.asc

[abdulazee...@gmail.com]

Hi Did you get this resolved? I'm having the same issue

[Denis Fayruzov]

Hi, 

as far as I remember I switched to iPXE and managed them to boot: 

:coreos

set base-url http://stable.release.core-os.net/amd64-usr/current

kernel ${base-url}/coreos_production_pxe.vmlinuz coreos.autologin console=ttyS1,115200n8 cloud-config-url=http://matchbox/cloud coreos.autologin console=tty0

initrd ${base-url}/coreos_production_pxe_image.cpio.gz

boot || goto failed

goto start

BR,

Denis