Le 18/02/2019 à 18:58, Tru Huynh a écrit :

On Mon, Feb 18, 2019 at 06:36:10PM +0100, Patrick Bégou wrote: ...

...

• PXE boot start and the desktop load the kernel and the inirtd file

• the kernel is booted and at this time the led of the network switch

goes down for the port used by the desktop.

• then I have a message "no carrier detected on interface em1"

• followed by " IPv6 ADDRCONF(NETDEV_CHANGED): em1: link become ready

(and of course the led goes back to green on the switch)

just add "nicdelay=50 linksleep=50" to your kickstart script (adapt the 50 secondes to your needs)

Cheers

Tru

I Tru,

Thanks for your answer. This morning I spent a long time testing but it is still not working. "Googling" with  *nicdelay* show me some interesting documentations about anaconda. I also try:

*ipv6.disable=1*, do disable the IPv6 setup, and *inst.waitfornet=30* options.

However when the kernel boot I never get any DHCP request to my server. I can see the DHCP request at PXE boot time, and it works, but none after dowlnoading vmlinuz and initrd.img.

The only solution was to provide the network setup the grub.cfg file and a waitfornet instruction:

set timeout=5     menuentry 'Centos7-UEFI-most-most8pc19' {     linuxefi centos74/vmlinuz *inst.waitfornet=30 ipv6.disable=1 ip=**/<desktop IP>/**::**/<gateway>/**:**/<mask>/**:**/<hostname>/**:em1:none *inst.repo=http:///<server IP>//centos74 inst.ks=http:///<server IP>//Kickstart7/optiplex9030_uefi_fr.cfg devfs=nomount     echo "Loading centos74/initrd.img"     initrdefi centos74/initrd.img     echo "Booting installation kernel" }

Realy strange as I boot and autoinstall many other PC from these server. I also notice that network is set to 100MbFD even if I have Gigabit switch. May be something related to the e1000 driver provided at boot time.

Patrick