FC15 TFTP open timeout.

[Francis SOUYRI]

Hello,

	Nobody have the same problem then me ?

Best regards.

Francis

On 09/15/2011 12:33 PM, Francis SOUYRI wrote:
> Hello,
>
> 	A missing information, there is no problem with a file with a size of 0
> (I suppose with a size less than 1 block).
>
> Best regards.
>
> Francis
>
> On 09/14/2011 02:41 PM, Francis SOUYRI wrote:
>> Hello,
>>
>> I have this:
>>
>> Sep 14 14:38:19 tauntauns1 systemd[1]: tftp.service failed to run
>> 'start' task: Invalid argument
>> Sep 14 14:38:19 tauntauns1 systemd[1]: Unit tftp.service entered failed
>> state.
>> Sep 14 14:38:19 tauntauns1 systemd[1]: tftp.service failed to run
>> 'start' task: Invalid argument
>> Sep 14 14:38:19 tauntauns1 systemd[1]: tftp.service failed to run
>> 'start' task: Invalid argument
>> Sep 14 14:38:19 tauntauns1 systemd[1]: tftp.service failed to run
>> 'start' task: Invalid argument
>> Sep 14 14:38:19 tauntauns1 systemd[1]: tftp.service failed to run
>> 'start' task: Invalid argument
>> Sep 14 14:38:19 tauntauns1 systemd[1]: tftp.service failed to run
>> 'start' task: Invalid argument
>> Sep 14 14:38:19 tauntauns1 systemd[1]: tftp.service start request
>> repeated too quickly, refusing to start.
>>
>> Best regards.
>>
>> Francis
>>
>> On 09/14/2011 01:17 PM, "Jóhann B. Guðmundsson" wrote:
>>> On 09/12/2011 10:10 AM, Francis SOUYRI wrote:
>>>> Hello,
>>>>
>>>> Since I upgrade our cluster from the FC14 to the FC15 I have a bad
>>>> problem with the tftpd server.
>>>> I have some thinstations booting over the network after the update the
>>>> boot failed with the message "TFTP open timeout". I have also the
>>>> problem with a fresh FC15 install.
>>>>
>>>> I use an "ip alias" (cluster address) for the tftp server (next-server
>>>> in the dhcp), when I use the real ip I have no problem.
>>>>
>>>> I do some test from a fc15 tftp client, to an "ip alias" I have this:
>>>
>>> I just finished converting the xinetd file to native systemd ones and
>>> was wondering if you would be willing to test that for us.
>>>
>>> Just download the attached files in 737212 ( tftp.socket and tftp.service ).
>>>
>>> Move/copy them into /etc/systemd/system or /lib/systemd/system directory
>>> and restart the daemoni ( systemctl daemon-reload )
>>>
>>> Disable the xinetd based tftp service and restart xinetd
>>>
>>> Start the tftp.socket ( systemctl start tftp.socket use systemctl status
>>> tftp.socket to see if it's running ) then test tftp.
>>>
>>>
>>> TFTP open timeout ( PXE-E32 ) happens when the PXE client was able to
>>> get a DHCP address and a boot file name, but timed out when attempting
>>> to download the boot file. Hence check Permission+Firewall+Networking
>>> and I should mention there can be problems with the physical network
>>> card as well so check bios/firmware etc..
>>>
>>>
>>> Btw you can test tftp by doing..
>>>
>>> #touch /var/lib/tftpboot/my-test-file.txt
>>> #chmod 777 /var/lib/tftpboot/my-test-file.txt
>>>
>>> #tftp $IP
>>>
>>> tftp>   put my-test-file.txt
>>>
>>> tftp>   quit
>>>
>>> Note if there is nothing comming to the logs or if you want to increase
>>> the verbosity of tftp you might need to add "StandardOutput=syslog" and
>>> "StandardError=syslog" to the [Service] section of the service file (
>>> tftp.service ) so it looks like this...
>>>
>>> [Unit]
>>> Description=Tftp Server After=local-fs.target
>>>
>>> [Service]
>>> ExecStart=/usr/sbin/in.tftpd -v -v -s /var/lib/tftpboot
>>> StandardOutput=syslog
>>> StandardError=syslog
>>> StandardInput=socket
>>>
>>>
>>> It would be good to get some feedback how this is working in an actual
>>> test/production setup
>>>
>>> Thanks
>>>
>>> JBG
>>
>