Updating to SVN 3121... Setting up and starting TFTP and PXE Servers...Failed!
-
[quote=“Berend de Boer, post: 44097, member: 28367”]What if you start it? “service tftpd-hpa start”[/quote]
[CODE]root@fogserver:~# service tftpd-hpa start
Starting HPA’s tftpd: in.tftpdroot@fogserver:~# service tftpd-hpa status
in.tftpd is not running … failed!
[/CODE] -
[quote=“Wayne Workman, post: 44096, member: 28155”]TFTP shouldn’t have a dependency on dnsmasq in order to run. (someone correct me if I’m wrong?)
Have you tried to restart TFTP?
What’s in the config file?[/quote]
/etc/default/tftpd-hpa
FOG Modified version
TFTP_USERNAME=“root”
TFTP_DIRECTORY=“/tftpboot”
TFTP_ADDRESS=“0.0.0.0:69”
TFTP_OPTIONS=“-s” -
[quote=“Jeremy Bailey, post: 44099, member: 12244”]/etc/default/tftpd-hpa
FOG Modified version
TFTP_USERNAME=“root”
TFTP_DIRECTORY=“/tftpboot”
TFTP_ADDRESS=“0.0.0.0:69”
TFTP_OPTIONS=“-s”[/quote]Simple question, did you edit the IP address to 0.0.0.0 just to hide it, or is that actually what was there?
-
That’s actually what was there.
-
[quote=“Jeremy Bailey, post: 44101, member: 12244”]That’s actually what was there.[/quote]
Just to back up for a moment (using the KISS method), if you go to the web GUI (you can still get to it, right?) and go to FOG settings, and go to the spot in there for TFTP settings, what do you see? Is it all correct?
What is inside of here?
[CODE]/etc/inetd.conf[/CODE]Also, try adding this line to /etc/default/tftpd-hpa
[CODE]RUN_DAEMON=“yes”[/CODE]First, backup what you have like this:
[CODE]sudo cp /etc/default/tftpd-hpa /etc/default/tftpd-hpa_backup[/CODE]You could add the “run daemon” line with an ECHO command, or you could use a graphical text editor. I think Debian comes with something called nedit. It’s up to you.
The ECHO command would be like this:
(make careful note of the beginning & ending single quotes, and the double quotes around yes)
[CODE]sudo echo ‘RUN_DAEMON=“yes”’ >> /etc/default/tftpd-hpa [/CODE]When you’re done, it should look like this:
[CODE]FOG Modified version
TFTP_USERNAME=“root”
TFTP_DIRECTORY=“/tftpboot”
TFTP_ADDRESS=“0.0.0.0:69”
TFTP_OPTIONS=“-s”
RUN_DAEMON=“yes”
[/CODE]Use the CAT command to confirm:
[CODE]sudo cat /etc/default/tftpd-hpa[/CODE]Then restart the service with:
[CODE]service tftpd-hpa restart[/CODE]Let us know what it says. If it makes it worse, you can just remove the line, or restore the backup…
-
From what I can see the config posted by Jeremy is perfectly fine. Maybe there is an old zombie of tftpd still running and listening on the port.
[CODE]sudo netstat -antup | grep “:69”[/CODE]
Kill any process you see before trying to start (service tftpd-hpa start) it again.If this does not help or if you don’t have any zombie processes you might try starting it in “debug mode”:
[CODE]sudo /usr/sbin/in.tftpd --foreground -vvv -s /tftpboot[/CODE]Then try tftp from another console:
[CODE]tftp 127.0.0.1
tftp> get default.ipxe
tftp>[/CODE]You should see something like this in the logs:
[CODE]sudo tail /var/log/daemon.log
…
… in.tftpd[26418]: RRQ from 127.0.0.1 filename default.ipxe[/CODE] -
Yeah, do what Uncle Frank said. The guy is a hero.
-
[quote=“Wayne Workman, post: 44102, member: 28155”]Just to back up for a moment (using the KISS method), if you go to the web GUI (you can still get to it, right?) and go to FOG settings, and go to the spot in there for TFTP settings, what do you see? Is it all correct?
What is inside of here?
[CODE]/etc/inetd.conf[/CODE]Also, try adding this line to /etc/default/tftpd-hpa
[CODE]RUN_DAEMON=“yes”[/CODE]First, backup what you have like this:
[CODE]sudo cp /etc/default/tftpd-hpa /etc/default/tftpd-hpa_backup[/CODE]You could add the “run daemon” line with an ECHO command, or you could use a graphical text editor. I think Debian comes with something called nedit. It’s up to you.
The ECHO command would be like this:
(make careful note of the beginning & ending single quotes, and the double quotes around yes)
[CODE]sudo echo ‘RUN_DAEMON=“yes”’ >> /etc/default/tftpd-hpa [/CODE]When you’re done, it should look like this:
[CODE]FOG Modified version
TFTP_USERNAME=“root”
TFTP_DIRECTORY=“/tftpboot”
TFTP_ADDRESS=“0.0.0.0:69”
TFTP_OPTIONS=“-s”
RUN_DAEMON=“yes”
[/CODE]Use the CAT command to confirm:
[CODE]sudo cat /etc/default/tftpd-hpa[/CODE]Then restart the service with:
[CODE]service tftpd-hpa restart[/CODE]Let us know what it says. If it makes it worse, you can just remove the line, or restore the backup…[/quote]
I finally got time to work on this again. Here’s a few observations. The same issue still exists in svn 3389.
inetd.conf does not exist.
I tried adding the RUN_DAEMON=“yes” with no success.
Here’s a quick look at my services. This is currently running 1.2.0, pre-install of the SVN. Is there any way to skip over the tftp and pxe install on the update?
service --status-all
[ ? ] FOGImageReplicator
[ ? ] FOGMulticastManager
[ ? ] FOGScheduler
[ + ] acpid
[ ? ] alsa-utils
[ - ] anacron
[ + ] apache2
[ + ] atd
[ + ] avahi-daemon
[ ? ] binfmt-support
[ + ] bluetooth
[ - ] bootlogs
[ ? ] bootmisc.sh
[ ? ] checkfs.sh
[ ? ] checkroot-bootclean.sh
[ - ] checkroot.sh
[ + ] clamav-freshclam
[ - ] console-setup
[ + ] cron
[ + ] dbus
[ + ] dnsmasq
[ + ] exim4
[ + ] gdm3
[ ? ] hdparm
[ - ] hostname.sh
[ ? ] hwclock.sh
[ ? ] initiscsi
[ - ] kbd
[ - ] keyboard-setup
[ ? ] killprocs
[ ? ] kmod
[ - ] lvm2
[ - ] minissdpd
[ - ] motd
[ ? ] mountall-bootclean.sh
[ ? ] mountall.sh
[ ? ] mountdevsubfs.sh
[ ? ] mountkernfs.sh
[ ? ] mountnfs-bootclean.sh
[ ? ] mountnfs.sh
[ ? ] mpt-statusd
[ ? ] mtab.sh
[ ? ] mysql
[ + ] network-manager
[ ? ] networking
[ + ] nfs-common
[ + ] nfs-kernel-server
[ + ] open-iscsi
[ ? ] pppd-dns
[ - ] procps
[ + ] pulseaudio
[ ? ] rc.local
[ - ] rmnologin
[ + ] rpcbind
[ - ] rsync
[ + ] rsyslog
[ + ] saned
[ ? ] sendsigs
[ ? ] speech-dispatcher
[ + ] ssh
[ - ] sudo
[ - ] tftpd-hpa
[ + ] udev
[ ? ] udev-mtab
[ ? ] umountfs
[ ? ] umountiscsi.sh
[ ? ] umountnfs.sh
[ ? ] umountroot
[ - ] unattended-upgrades
[ - ] urandom
[ + ] vmware-tools
[ + ] vsftpd
[ - ] x11-common
[ ? ] xinetd -
So I have figured out some of my issue. It does actually have to do with DNSMASQ.
When I first configured my fog server, I had decided to use the “enable-tftp” on the ltsp.conf file in /etc/dnsmasq.d. After symlinking the file for undionly.0, PXE worked great and I never upgraded from 1.2.0.
Until the surface pro 3 came along! Obviously I have a network with mixed devices and I can’t hand out just one boot file anymore. I can change the bootfile on a per host basis on my dhcp server, so I have done that for testing purposes.
Commenting out with #enable-tftp allowed for tftpd-hpa to listen on port 69 once again. Ran the upgrade to subversion and successfully ran an installer. I’m now at 3510. The problem is that the only way I can push a bootfile is by once again enabling tftp from the DNSMASQ service and stopping tftpd-hpa. Both cannot listen on the same port, they error out.
I need dhcp to point my surface pros to snp.efi in order for it to work, but I believe the only way that is going to successfully work is with tftpd-hpa. But the only way I can receive a bootfile seems to be with DNSMASQ.
Do you have any guidance?
Thanks again for all of your help.
-
Out of pure curiosity, what version of Debian are you using?
-
Debian version 7.8
-
I’d just like to state that this issue has been solved.
In summary, the installer will fail if you use DNSMASQ’s built-in tftp server instead of tftpd-hpa.
Thank you for all of your help!
Please close
