F23 PXE strange failure

Tuesday, 19 January 2016

I'm working with a new PXE configuration file which may certainly 
contain some errors. Usually, it terminates with a "pane dead". But 
after the latest modification, it terminated by going to the Dracut 
emergency shell. In itself, that might help me, even though thus far I 
haven't found my error in the logs. But I did find something that seems 
quite strange to me and would like any available feedback.

Inside the Dracut shell, I ran the journalctl command. As this was a 
server in a remote data center, I couldn't easily attach a thumb drive 
and save the result. But here are the last few lines output.

Jan 19 20:35:52 localhost dracut-initqueue[812]: Starting dhcp for 
interface eth0
Jan 19 20:35:52 localhost dracut-initqueue[812]: dhcp: PREINIT eth0 up
Jan 19 20:35:52 localhost dhclient[1041]: Created duid 
\000\001\000\001\03612\250\260\203\376\341\211.
Jan 19 20:35:52 localhost dhclient[1041]: DHCPDISCOVER on eth0 to 
255.255.255.255 port 67 interval 8 (xid=0x45a50c20)
Jan 19 20:36:00 localhost dhclient[1041]: DHCPDISCOVER on eth0 to 
255.255.255.255 port 67 interval 18 (xid=0x45a50c20)
Jan 19 20:36:18 localhost dhclient[1041]: DHCPDISCOVER on eth0 to 
255.255.255.255 port 67 interval 8 (xid=0x45a50c20)
Jan 19 20:36:26 localhost dhclient[1041]: DHCPDISCOVER on eth0 to 
255.255.255.255 port 67 interval 15 (xid=0x45a50c20)
Jan 19 20:36:41 localhost dhclient[1041]: DHCPDISCOVER on eth0 to 
255.255.255.255 port 67 interval 12 (xid=0x45a50c20)
Jan 19 20:36:53 localhost dhclient[1041]: No DHCPOFFERS received.
Jan 19 20:36:53 localhost cracut-initqueue[812]: dhcp failed
Jan 19 20:36:59 localhost kernel: random: nonblocking pool is initialized
Jan 19 20:40:01 localhost dracut-initqueue[812]: Warning: Could not boot.
Jan 19 20:40:01 localhost systemd[1]: Received SIGRTMIN+20 from PID 823 
(plymouthd).
Jan 19 20:40:01 localhost dracut-initqueue[812]: Warning: /dev/root does 
not exi
st
Jan 19 20:40:01 localhost systemd[1]: Starting Dracut Emergency Shell...

Ok, so the PXE startup doesn't think it was able to get an IP address. 
But when I look at the DHCP server, it did get an address! The times 
from journalctl are UTC and those from the DHCP log are local PST 
(-0800) but the times otherwise match up quite well. Here's what the 
DHCP log shows.

Jan 19 12:34:52 us02linuxdhcp01p dhcpd[8500]: DHCPDISCOVER from 
b0:83:fe:e1:cc:89 via 10.76.148.1
Jan 19 12:34:53 us02linuxdhcp01p dhcpd[8500]: DHCPOFFER on 10.76.148.243 
to b0:83:fe:e1:cc:89 via 10.76.148.1
Jan 19 12:34:57 us02linuxdhcp01p dhcpd[8500]: DHCPREQUEST for 
10.76.148.243 (10.76.140.72) from b0:83:fe:e1:cc:89 via 10.76.148.1
Jan 19 12:34:57 us02linuxdhcp01p dhcpd[8500]: DHCPACK on 10.76.148.243 
to b0:83:fe:e1:cc:89 via 10.76.148.1
Jan 19 12:36:01 us02linuxdhcp01p dhcpd[8500]: DHCPDISCOVER from 
b0:83:fe:e1:cc:89 via 10.76.148.1: network 10.76.148.0/23: no free leases
Jan 19 12:36:18 us02linuxdhcp01p dhcpd[8500]: DHCPDISCOVER from 
b0:83:fe:e1:cc:89 via 10.76.148.1: network 10.76.148.0/23: no free leases
Jan 19 12:36:26 us02linuxdhcp01p dhcpd[8500]: DHCPDISCOVER from 
b0:83:fe:e1:cc:89 via 10.76.148.1: network 10.76.148.0/23: no free leases
Jan 19 12:36:42 us02linuxdhcp01p dhcpd[8500]: DHCPDISCOVER from 
b0:83:fe:e1:cc:89 via 10.76.148.1: network 10.76.148.0/23: no free leases

Not only was an OFFER sent in response to the first REQUEST, it appears 
that the OFFER was accepted by the client. Still the client sent four 
additional REQUESTs and didn't record the result of the acceptance. How 
can this be?
-- 
Dave Close