1

I have a FreeBSD server on which I would like to run a TFTP server (for PXE). My daemon of choice is tftp-hpa and I have it configured to serve from /tftpboot properly. It works unless I mount /tftpboot via NFS. Here is some more information...

  • sockstat -4l confirms in.tftpd is running on *:69 (which is correct)
  • telnet obviously doesn't work (UDP) but when I try to grab a file that exists, I get a timeout.
  • There's no loot in /var/log/messages
  • When I request a file that does not exist, I get the proper error

Any ideas?

Alex Hirzel
  • 186
  • 6
  • I do the same thing at work, with tftpd-hpa, and it works fine. Albeit on linux, not freebsd. – Tom O'Connor Dec 28 '11 at 23:19
  • Try adding -v to the tftp server's launch arguments. This may give you a bit more information to debug this. – devicenull Dec 29 '11 at 01:24
  • @devicenull, that led me to believe it was a problem outside of the TFTP server. Verbose logging showed all requests properly (including retries) but gave no evidence of a response. – Alex Hirzel Dec 29 '11 at 19:46

2 Answers2

1

If your NFS mount is exported read-only (regardless of whether it is mounted rw or ro), this seems to cause tftpd-hpa to exhibit this behavior. While I’ve no idea why, exporting rw resolves it.

Andrew Marshall
  • 432
  • 4
  • 13
0

I did not end up resolving this. I opted to serve TFTP from the file server (which exports /tftpboot). In addition to actually working, it's a lot faster. If anyone wants help debugging this issue in the future, feel free to contact me.

Alex Hirzel
  • 186
  • 6