DHCP and e1000

mail at kwerner.net mail at kwerner.net
Sun Dec 10 12:37:22 CET 2006 

Hello,

thanks very much for your patience and cooperation.

> Do you maked a "fai-chboot -IB 192.168.7.113" on the server?
> 
> Give me please the output of "ls -l /srv/tftp/fai/pxelinux.cfg/".

faitest2:~# ls -l /srv/tftp/fai/pxelinux.cfg/
insgesamt 2
-rw-r--r-- 1 root root 269 2006-12-09 12:40 C0A80771	<--- dualxeon
-rw-r--r-- 1 root root 266 2006-12-09 14:15 C0A80773	<--- amdk6

and this what I did (at this stage of debugging I do not want to perform
an installation, sysinfo is sufficient for the time being):

Client DUALXEON (this one is unable to perform a successful mount)

faitest2:~# fai-chboot -FSv dualxeon
append parameters: ip=dhcp FAI_ACTION=sysinfo
FAI_FLAGS=verbose,sshd,createvt
Booting kernel vmlinuz-install
rootfs is /dev/nfs nfsroot=/srv/fai/nfsroot,v3,tcp,rsize=32768,wsize=32768
dualxeon has 192.168.7.113 in hex C0A80771
Writing file /srv/tftp/fai/pxelinux.cfg/C0A80771 for dualxeon

DHCP config for this host:

host dualxeon {
  hardware ethernet 00:e0:81:72:04:bb;
  fixed-address 192.168.7.113;
}

BIND9 config:

dualxeon                A       192.168.7.113
113                     PTR     dualxeon.urandom.lan.


and

Client AMDK6 (this one has no problem at all)

faitest2:~# fai-chboot -FSv amdk6
append parameters: ip=dhcp FAI_ACTION=sysinfo
FAI_FLAGS=verbose,sshd,createvt
Booting kernel vmlinuz-install
rootfs is /dev/nfs nfsroot=/srv/fai/nfsroot,v3,tcp,rsize=32768,wsize=32768
amdk6 has 192.168.7.115 in hex C0A80773
Writing file /srv/tftp/fai/pxelinux.cfg/C0A80773 for amdk6

DHCP config for this host:

host amdk6 {
  hardware ethernet 00:60:08:69:90:c4;
  fixed-address 192.168.7.115;
}

BIND9 config:

amdk6                   A       192.168.7.115
115                     PTR     amdk6.urandom.lan.

DHCP configuration for my subnet:

subnet 192.168.7.0 netmask 255.255.255.0 {
        option broadcast-address 192.168.7.255;
        range 192.168.7.200 192.168.7.250;
        option routers 192.168.7.1;
        option domain-name "urandom.lan";
        option domain-name-servers 192.168.7.76, 192.168.7.1;
        default-lease-time 600;
        max-lease-time 7200;
        option time-servers 192.168.7.76;
        option ntp-servers 192.168.7.76;
        server-name "192.168.7.181";  <--- faitest2 (my FAISERVER)
        next-server 192.168.7.181;
        option root-path
"/srv/fai/nfsroot,rsize=8192,wsize=8192,acregmin=1800,acregmax=1800,acdirmin=1800,acdirmax=1800";
        ddns-domainname "urandom.lan";
        filename "/pxelinux.0";
        get-lease-hostnames true;
        use-host-decl-names on;
}


> Dec  9 18:28:25 amd1800 in.tftpd[4895]: sending NAK (1, File not found)
>> to 192.168.7.113

This not a bug, in fact it is a feature of the TFTPD. At first TFTPD
tries to load a file that is named after the clients MAC address. Such a
file does not exist on my server, so he prints a "File not found"
message to syslog (tftpd-hpa is not as verbose as atftpd is).
At the second stage the TFTPD looks for a file that is named after the
clients IP address in HEX format. These files exist for both of the
concerned clients and they are served. If this was not the case, they
would not be able to load a kernel at all. But both do.
The problem I have, does not occur in this phase, as you can see below.

Output from syslog:

Client DUALXEON (this one is unable to perform a successful mount)

Dec  9 13:36:04 faitest2 atftpd[3502]: Serving
/pxelinux.cfg/01-00-e0-81-72-04-bb to 192.168.7.113:57089
Dec  9 13:36:04 faitest2 atftpd[3502]: File
/srv/tftp/fai//pxelinux.cfg/01-00-e0-81-72-04-bb not found
Dec  9 13:36:04 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:04 faitest2 atftpd[3502]: Serving /pxelinux.cfg/C0A80771 to
192.168.7.113:57090


Client AMDK6 (this one has no problem at all)

Dec  9 12:53:25 faitest2 atftpd[3501]: Serving
/pxelinux.cfg/01-00-60-08-69-90-c4 to 192.168.7.115:57089
Dec  9 12:53:25 faitest2 atftpd[3501]: File
/srv/tftp/fai//pxelinux.cfg/01-00-60-08-69-90-c4 not found
Dec  9 12:53:25 faitest2 atftpd[3501]: Server thread exiting
Dec  9 12:53:25 faitest2 atftpd[3501]: Serving /pxelinux.cfg/C0A80773 to
192.168.7.115:57090


And that is what really puzzles me. The first contact of any client to
the DHCPD is successfull. Both clients send out requests to the server
and get assigned an IP address, gateway, the path to their PXE
configuration and so on. After that both clients load the install kernel.

Enter the problem.
And here it comes: The one client (amdk6) -actually the install kernel
that is configured to use DHCP to find its NFSROOT - sends out another
request to the DHCPD that tells him, where to look for
(/srv/fai/nfsroot)  and mounts the NFSROOT:

Dec  9 12:53:25 faitest2 atftpd[3501]: Serving /pxelinux.0 to
192.168.7.115:2070
Dec  9 12:53:25 faitest2 atftpd[3501]: tsize option -> 13324
Dec  9 12:53:25 faitest2 atftpd[3501]: Server thread exiting
Dec  9 12:53:25 faitest2 atftpd[3501]: Serving /pxelinux.0 to
192.168.7.115:2071
Dec  9 12:53:25 faitest2 atftpd[3501]: blksize option -> 1456
Dec  9 12:53:25 faitest2 atftpd[3501]: Server thread exiting
Dec  9 12:53:25 faitest2 atftpd[3501]: Serving
/pxelinux.cfg/01-00-60-08-69-90-c4 to 192.168.7.115:57089
Dec  9 12:53:25 faitest2 atftpd[3501]: File
/srv/tftp/fai//pxelinux.cfg/01-00-60-08-69-90-c4 not found
Dec  9 12:53:25 faitest2 atftpd[3501]: Server thread exiting
Dec  9 12:53:25 faitest2 atftpd[3501]: Serving /pxelinux.cfg/C0A80773 to
192.168.7.115:57090
Dec  9 12:53:25 faitest2 atftpd[3501]: tsize option -> 266
Dec  9 12:53:25 faitest2 atftpd[3501]: blksize option -> 1440
Dec  9 12:53:25 faitest2 atftpd[3501]: Server thread exiting
Dec  9 12:53:25 faitest2 atftpd[3501]: Serving /vmlinuz-install to
192.168.7.115:57091
Dec  9 12:53:25 faitest2 atftpd[3501]: tsize option -> 1579042
Dec  9 12:53:25 faitest2 atftpd[3501]: blksize option -> 1440
Dec  9 12:53:27 faitest2 atftpd[3501]: Server thread exiting
Dec  9 12:53:30 faitest2 dhcpd: DHCPDISCOVER from 00:60:08:69:90:c4 via eth0
Dec  9 12:53:30 faitest2 dhcpd: DHCPOFFER on 192.168.7.115 to
00:60:08:69:90:c4 via eth0
Dec  9 12:53:30 faitest2 dhcpd: DHCPREQUEST for 192.168.7.115
(192.168.7.181) from 00:60:08:69:90:c4 via eth0
Dec  9 12:53:30 faitest2 dhcpd: DHCPACK on 192.168.7.115 to
00:60:08:69:90:c4 via eth0
Dec  9 12:53:30 faitest2 mountd[3597]: authenticated mount request from
amdk6.urandom.lan:780 for /srv/fai/nfsroot (/srv/fai/nfsroot)


The system in question (dualxeon) also sends out requests. These
requests reach the DHCPD and he answers it. But these answers are not
acknowledged by my client, so there is no way he can mount the NFSROOT.

Dec  9 13:35:58 faitest2 atftpd[3502]: Serving /pxelinux.0 to
192.168.7.113:2070
Dec  9 13:35:58 faitest2 atftpd[3502]: tsize option -> 13324
Dec  9 13:36:00 faitest2 atftpd[3502]: Serving /pxelinux.0 to
192.168.7.113:2071
Dec  9 13:36:00 faitest2 atftpd[3502]: tsize option -> 13324
Dec  9 13:36:03 faitest2 atftpd[3502]: timeout: retrying...
Dec  9 13:36:04 faitest2 atftpd[3502]: Serving /pxelinux.0 to
192.168.7.113:2072
Dec  9 13:36:04 faitest2 atftpd[3502]: tsize option -> 13324
Dec  9 13:36:04 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:04 faitest2 atftpd[3502]: Serving /pxelinux.0 to
192.168.7.113:2073
Dec  9 13:36:04 faitest2 atftpd[3502]: blksize option -> 1456
Dec  9 13:36:04 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:04 faitest2 atftpd[3502]: Serving
/pxelinux.cfg/01-00-e0-81-72-04-bb to 192.168.7.113:57089
Dec  9 13:36:04 faitest2 atftpd[3502]: File
/srv/tftp/fai//pxelinux.cfg/01-00-e0-81-72-04-bb not found
Dec  9 13:36:04 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:04 faitest2 atftpd[3502]: Serving /pxelinux.cfg/C0A80771 to
192.168.7.113:57090
Dec  9 13:36:04 faitest2 atftpd[3502]: tsize option -> 269
Dec  9 13:36:04 faitest2 atftpd[3502]: blksize option -> 1440
Dec  9 13:36:04 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:04 faitest2 atftpd[3502]: Serving /vmlinuz-install to
192.168.7.113:57091
Dec  9 13:36:04 faitest2 atftpd[3502]: tsize option -> 1579042
Dec  9 13:36:04 faitest2 atftpd[3502]: blksize option -> 1440
Dec  9 13:36:05 faitest2 atftpd[3502]: timeout: retrying...	<----
Dec  9 13:36:13 faitest2 last message repeated 4 times
Dec  9 13:36:15 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:15 faitest2 atftpd[3502]: timeout: retrying...	<----
Dec  9 13:36:25 faitest2 last message repeated 4 times
Dec  9 13:36:28 faitest2 atftpd[3502]: client (192.168.7.113) not
responding							<----
Dec  9 13:36:28 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:30 faitest2 atftpd[3502]: client (192.168.7.113) not
responding							<----
Dec  9 13:36:30 faitest2 atftpd[3502]: Server thread exiting
Dec  9 13:36:31 faitest2 dhcpd: DHCPDISCOVER from 00:e0:81:72:04:bb via eth0
Dec  9 13:36:31 faitest2 dhcpd: DHCPOFFER on 192.168.7.113 to
00:e0:81:72:04:bb via eth0
(The last two lines are being repeated until the client is shut down, no
"DHCPACK" occurs.)


I really appreciate your help and am sorry for my verbose (even
redundant) description of my problem, but I want to give as much
information as possible. Maybe I am in the wrong place to look for
answers after all, because tis symptom may not be connected to FAI in
the first place. I was hoping someone with a similar setup may have run
into the same mess and found a solution to it.

So, thanks again.

K. Werner

More information about the linux-fai mailing list