PXE Boot Problems / TFTP Problem

[Fogbrain]

Recieving transfer timeout for following command
tftp -v <serverIP> -m binary -c get undionly.kpxe

Using TCPDUMP and can see the following
dropped privs to tcpdump
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on any, link-type LINUX_SLL (Linux cooked v1), capture size 262144 bytes
16:41:09.411021 IP <sourceIP>.57485 > <fogIP>.69: 22 RRQ “undionly.kpxe” octet
16:41:14.412414 IP <sourceIP>.57485 > <fogIP>.69: 22 RRQ “undionly.kpxe” octet
16:41:19.413627 IP <sourceIP>.57485 > <fogIP>.69: 22 RRQ “undionly.kpxe” octet
16:41:24.414186 IP <sourceIP>.57485 > <fogIP>.69: 22 RRQ “undionly.kpxe” octet
16:41:29.414997 IP <sourceIP>.57485 > <fogIP>.69: 22 RRQ “undionly.kpxe” octet
5 packets captured
5 packets received by filter
0 packets dropped by kernel

Any idea’s/help would be greatly appreciated.

[george1421]

@fogbrain I’d like to see the complete pcap as outlined in this tutorial. https://forums.fogproject.org/topic/9673/when-dhcp-pxe-booting-process-goes-bad-and-you-have-no-clue

Post it to a file share site, post it as public read with link, and either post the link here or use FOG chat and send me the link so I can look at the complete process.

If you suspect tftp is not working, from a windows computer load the tftp client program, turn off the windows firewall and see if you can get the file with tftp get command.

[Fogbrain]

@george1421

Hi George,

Thanks for the reply, It was an issue with our firewall rules, I’ve resolved that now, But now I’m receiving the following when trying to do a full registration of the host

“Either DHCP failed or we were unable to access http://<IP>/fog//index.php for connection testing for connection testing”

I’ve looked at the post here : https://forums.fogproject.org/topic/13478/either-dhcp-failed-or-we-were-unable-to-access-http-ip-fog-index-php-for-connection-testing?lang=en-US

The user tlehrian recommends running a different spanning-tree than the default, we are currently using RSPT on the switch the client and FOG server is connected to.

SW Port config is:
switchport mode access
no cdp enable
spanning-tree portfast

[Fogbrain]

@george1421
https://mega.nz/file/3F9lBQgY#JN5e-ePcdE9BQxpyai3h7sQAYXJFJjImmUKRWECDBKU

[george1421]

@fogbrain That message has 2 causes.

1. It can’t get an IP address
2. It can’t make a http call to the fog server.

If you watch it boot you should see it get an IP address, if yes then that rules out #1. The second is the FOS Engine will attempt to reach the fog server (as you posted in the error above). If it can’t it will start the sequence over again of querying for dhcp then attempting to contact the fog server. It should do this 3 times then give up. You need to test from the same subnet as the pxe booting computer if you can reach the url you defined in your original post with a different computer on the same subnet as the pxe booting computer.

I don’t think its a spanning tree issue right now. That would usually (but not always) show up in the FOG iPXE menu not loading. You do need to have your switch ports in portfast mode. You can test to see if its spanning tree issue by placing a dumb (read cheap, unmanaged) switch between the building switch and the pxe booting computer. That cheap switch typically would not have spanning tree plus it keeps the building switch port from winking as the pxe booting computer starts up.