Thank you für your reply.

Yes, that’s what the log file state’s. The change of the hostname work’s fine, but joining the domain fails (DELL optiplex 790). With the same image on another machine (optiplex 780) the domain join work’s fine.

Could it be, that the Domain joining process start’s too soon when other Network processes (GPOs, …) are still working?

I’ ve different images (all Windows 7) working fine. But with one image there occurs an error while joining the domain (Hostname changer … error 1219).
If I deploy a .cmd snapin to the Client with the netdom join … command, everything works fine.

Thanks for your reply. The problem appears during unicast.
But It wasn’t a switch problem but a problem with the client Broadcom NIC driver in the kernel. I compiled a new kernel with a different NIC driver and everything works now.

I replaced my old switches (100mbit). Now there are HP Procurve2524G Switches (1000mbit). With the old switches everthing went fine. Now, when I deploy an image to a client, it hangs after ~100MB.
I use the newest kernel, I think the problem is the configuration of the switch?

Any ideas?

I think the reason is the Ubuntu start mechanism “Upstart”, where the daemons do not start step by step but almost simultaneously.
So tftpd-hpa starts “to early”. You can fix this by editing the /etc/rc.local and ad the line

(/bin/sleep 30 && /etc/init.d/htftpd-hpa restart)

to restart the daemon after 30 seconds

ullawolf1