TFTP and IPv4

[carles]

Dear OPSI comminity,

I am having a problem with the PXE TFTP boot, on the clients:

PXE-E32: TFTP open timeout

I tried with :

opsi-setup --set-rights /tftpboot

with no luck,

I checked the syslog on the opsi server and this is what i found:

Nov 29 15:20:14 pxe01 puppet-agent[959]: Finished catalog run in 57.58 seconds
Nov 29 15:20:34 pxe01 atftpd[1322]: connect: Address family not supported by protocol
Nov 29 15:21:46 pxe01 atftpd[1322]: connect: Address family not supported by protocol
Nov 29 15:23:34 pxe01 atftpd[1322]: connect: Address family not supported by protocol
Nov 29 15:25:58 pxe01 atftpd[1322]: connect: Address family not supported by protocol
Nov 29 15:30:58 pxe01 atftpd[1322]: atftpd terminating after 300 seconds
Nov 29 15:30:58 pxe01 atftpd[1322]: Main thread exiting

This log made me ask if maybe TFTP was trying to run with IPv6, so:

[root@pxe01]: / # lsof -i | grep tftp
xinetd 797 root 6u IPv6 3653 0t0 UDP *:tftp
in.tftpd 6068 nobody 0u IPv6 3653 0t0 UDP *:tftp
in.tftpd 6068 nobody 1u IPv6 3653 0t0 UDP *:tftp
in.tftpd 6068 nobody 2u IPv6 3653 0t0 UDP *:tftp

How is this possible? I didn't changed anything in the server more than normal security updates, any idea how I can try to solve this problem? I am not interested yet in ipv6... I tried what I know, that was disabling IPv6 in the server, but nothing changed.

Any help would be really appreciated, Thank you !

Carles

[ueluekmen]

Hi,

it looks like your are using inetd to start atftpd.

Please check you inetd.conf which option you have set for opsi-atftpd.

[carles]

Thanks for your answer,

This is my inetd.conf, I don't detect anything rare there:

#:BOOT: TFTP service is provided primarily for booting. Most sites
# run this only on machines acting as "boot servers."
tftp dgram udp wait nobody /usr/sbin/tcpd /usr/sbin/in.tftpd --tftpd-timeout 300 --retry-timeout 5 --mcast-port 1758 --mcast-addr 239.239.239.0-255 --mcast-ttl 1 --maxthread 100 --verbose=5 /tftpboot

Any idea?

[ueluekmen]

Ok, that looks not wrong.

Please try to modify your inetd.conf and set verbose=7. Kill all running in.tftpd processes and send kill -HUP to the inetd process. Try again and check syslog. Perhaps you see a liitle bit more, whats happening.

[carles]

Dear uelekmen, thanks for your answer again

This is my syslog after changing the verbose to 7 and restarting the proceses:


Dec 4 10:24:57 pxe01 xinetd[813]: Started working: 2 available services
Dec 4 10:24:57 pxe01 /etc/mysql/debian-start[818]: /usr/bin/mysql_upgrade: the '--basedir' option is always ignored
Dec 4 10:24:57 pxe01 /etc/mysql/debian-start[818]: Looking for 'mysql' as: /usr/bin/mysql
Dec 4 10:24:57 pxe01 /etc/mysql/debian-start[818]: Looking for 'mysqlcheck' as: /usr/bin/mysqlcheck
Dec 4 10:24:57 pxe01 /etc/mysql/debian-start[818]: This installation of MySQL is already upgraded to 5.1.72, use --force if you still need to run mysql_upgrade
Dec 4 10:24:57 pxe01 /etc/mysql/debian-start[832]: Checking for insecure root accounts.
Dec 4 10:24:57 pxe01 /etc/mysql/debian-start[836]: Triggering myisam-recover for all MyISAM tables
Dec 4 10:24:59 pxe01 ntpdate[393]: step time server 91.189.89.199 offset 1.543761 sec
Dec 4 10:25:01 pxe01 puppet-agent[1064]: Reopening log files
Dec 4 10:25:01 pxe01 puppet-agent[1064]: Starting Puppet client version 3.3.2
Dec 4 10:26:10 pxe01 puppet-agent[1084]: Finished catalog run in 53.11 seconds
Dec 4 10:26:47 pxe01 in.tftpd[2013]: connect from ::ffff:172.20.177.126 (::ffff:172.20.177.126)
Dec 4 10:26:47 pxe01 atftpd[2013]: Advanced Trivial FTP server started (0.7)
Dec 4 10:26:47 pxe01 atftpd[2013]: started by inetd
Dec 4 10:26:47 pxe01 atftpd[2013]: logging level: 7
Dec 4 10:26:47 pxe01 atftpd[2013]: directory: /tftpboot/
Dec 4 10:26:47 pxe01 atftpd[2013]: user: nobody.nogroup
Dec 4 10:26:47 pxe01 atftpd[2013]: log file: syslog
Dec 4 10:26:47 pxe01 atftpd[2013]: server timeout: 300
Dec 4 10:26:47 pxe01 atftpd[2013]: tftp retry timeout: 5
Dec 4 10:26:47 pxe01 atftpd[2013]: maximum number of thread: 100
Dec 4 10:26:47 pxe01 atftpd[2013]: option timeout: enabled
Dec 4 10:26:47 pxe01 atftpd[2013]: option tzise: enabled
Dec 4 10:26:47 pxe01 atftpd[2013]: option blksize: enabled
Dec 4 10:26:47 pxe01 atftpd[2013]: maximum blocksize: 0
Dec 4 10:26:47 pxe01 atftpd[2013]: option multicast: enabled
Dec 4 10:26:47 pxe01 atftpd[2013]: address range: 239.239.239.0-255
Dec 4 10:26:47 pxe01 atftpd[2013]: port range: 1758
Dec 4 10:26:47 pxe01 atftpd[2013]: connect: Address family not supported by protocol
Dec 4 10:26:47 pxe01 atftpd[2013]: Creating new socket: 172.20.180.81:49373
Dec 4 10:26:47 pxe01 atftpd[2013]: Server thread exiting
Dec 4 10:26:49 pxe01 atftpd[2013]: connect: Address family not supported by protocol
Dec 4 10:26:49 pxe01 atftpd[2013]: Creating new socket: 172.20.180.81:38607
Dec 4 10:26:49 pxe01 atftpd[2013]: Server thread exiting
Dec 4 10:26:53 pxe01 atftpd[2013]: connect: Address family not supported by protocol
Dec 4 10:26:53 pxe01 atftpd[2013]: Creating new socket: 172.20.180.81:34799
Dec 4 10:26:53 pxe01 atftpd[2013]: Server thread exiting
Dec 4 10:26:59 pxe01 atftpd[2013]: connect: Address family not supported by protocol
Dec 4 10:26:59 pxe01 atftpd[2013]: Creating new socket: 172.20.180.81:40811
Dec 4 10:26:59 pxe01 atftpd[2013]: Server thread exiting

[SisterOfMercy]

tftp dgram udp wait nobody /usr/sbin/tcpd /usr/sbin/in.tftpd --tftpd-timeout 300 --retry-timeout 5 --mcast-port 1758 --mcast-addr 239.239.239.0-255 --mcast-ttl 1 --maxthread 100 --verbose=5 /tftpboot

Change udp to udp4 in this configuration file and try again.

[carles]

when I try with udp4 in the inetd.conf file what appears in syslog is :

Dec 5 10:18:54 pxe01 xinetd[810]: inetd.conf - Protocol udp4 not in /etc/protocols [file=/etc/inetd.conf] [line=32]

[carles]

ok, I discovered the issue, first of all let me express my gratitude to everybody here, now i am going to explain what happened so maybe it can be useful for someone else:

My puppet master server was installing xinetd on the opsi server, which was needed by check_mk. When xinetd was installed it replaced the packet openbsd-inetd and everything stopped working.

In the place of creating new puppet recipe for the opsi server, what I have done is to move the inetd.conf to a xinetd format, in /etc/xinetd.d/tftpd

service tftp
{
socket_type = dgram
protocol = udp
wait = yes
user = root
server = /usr/sbin/in.tftpd
server_args = --tftpd-timeout 300 --retry-timeout 5 --mcast-port 1758 --mcast-addr 239.239.239.0-255 --mcast-ttl 1 --maxthread 100 --verbose=5 /tftpboot
disable = no
}


thank you very much