[K12OSN] Guidance on troubleshooting - new RHEL6 + LTSP won't PXE

[Shawn Wells]

Hi Josh, Warren,

	Thanks for the replies.  I'm just getting back online after the 
holidays here.  Comments in-line:


On 11/23/11 4:33 PM, Josh Malone wrote:
> On 23.11.2011 16:11, Shawn Wells wrote:
>
>
>> Per the instructions I configured ltspbr0 and tried to PXE boot the
>> VM which is acting as the thin client... no luck.  My config is just
>> two VMs for testing purposes.
>>
>> I receive the following error:
>> PXE-E51: No DHCP or proxyDHCP offers were received.
>
> <snip>
>
>> After failing to PXE boot I loaded an OS on the thin client and was
>> able to ping 172.31.100.254, so it doesn't appear to be a connectivity
>> issue.
>
> So, your clients in "fat mode" can get a DHCP address, but PXE clients
> cannot?


The "fat mode" clients I loaded an OS and statically assigned it an IP 
address within the 172.31.100.x range. It was able to ping the ltsp 
server at 172.31.100.254.


>
>
>> I went ahead and rebooted my server, then tried to PXE again.  No dice.
>>
>> I ensured DHCP is running and listening for connections:
>> # chkconfig --list | grep dhcp
>> ltsp-dhcp    0:off    1:off    2:on    3:on    4:on    5:on    6:off
>>
>> # service ltsp-dhcp status
>> dhcpd (pid 1920) is running...
>>
>> $ netstat -an | fgrep -w 67
>> udp    0    0.0.0.0:67    0.0.0.0:*
>
> If dhcpd is running, have you checked to make sure it's listening on the
> right network interface? Tcpdump may be your friend here.
>


Yes, I went as far as removing the other devices from the VM completely 
just to make sure of this!



>
>> I then ensured tftp was running and enabled
>
> Your problem is before TFTP.
>
> I'm sorry I'm not much help here. In my case, I'm not using the LTSP
> server as the DHCP or network router/nat; my clients are on my internal
> network just like any other host, receiving dhcp addresses from my
> network's existing DHCP server. I'm not completely sure I understand the
> ltspbr0 function.

Any response is welcome, especially those which validate that some of 
the testing I've done already were the right paths to head down.




> On 11/24/11 5:45 PM, Warren Togami Jr. wrote:
>> Hi Shawn,
>>
>> Your attempt is complicated substantially by running the LTSP server
>> within a VM.
>>
>> * The interface that you attach to ltspbr0 in the server VM needs to be
>> a secondary virtual eth device, not the primary that the VM uses to
>> reach the Internet.

Correct.  On my setup eth0 leads to the public internet, whereas eth1 is 
private.



>> * Then on the VM host, that secondary virtual eth device needs to attach
>> to an entirely separate bridge that probably doesn't already exist on
>> your system. This new bridge that you must create must not have any IP
>> address. It exists only for VM's to connect to each other.

I created a new virtual bridge (virbr1) which just allows communication 
between guests attached to it. So far so good.  When the LTSP-Server and 
another VM are attached to this they're able to ping each other, so it 
looks like the comms aspect is OK.


>> * Then the PXE client VM must connect to that new bridge.

For this I assigned the secondary interface of the server (for me, eth1) 
to virbr1.


>> If you set this up properly, then server VM's ltspbr0 connects to the
>> virt host bridge, and the PXE client VM can also connected to that same
>> bridge should be able to DHCP and PXE boot.

Still no luck on my side, even after reloading the LTSP Server and 
starting anew. I'll have access to physical hardware later this week to 
test things out on.

Also, it seems like many more people have had this deployed on RHEL5 
whereas I'm on RHEL6.  Any chance this could be causing issues?

Thanks!
Shawn