Can't PXE Boot [but the PXE request succeeded] - Error Tftp

[Ron Baduach]

Hi All,

It's a little bit complex issue, but hope you'll be able to assist with that .

Issue's Brief :

A blade is getting a PXE boot connection, but failed with an error :

TFTP cannot open connection

The issue, more detailed :

I'm having a UCS Director Bare Metal Agent (a VM), and there is a PXE server on it .

I'm succeeded to PXE VMs that have a NIC with the PXE Vlan [Native]

While trying to PXE boot a physical blade, it looks like the request is getting to the blade, but I'm getting the following error : TFTP cannot open connection [also see screenshot] .

The blade's Service Profile have a NIC with the PXE Vlan also [Native] .

 

UCS Director specialist told me that if I'm able to PXE boot on a VM, and on the physical blade it's not working yet, it should be some network issue that I'm missing somewhere .

Can you guide me what can I check in order to see if I have some issue ? do you have any clue what is going on ?

Again, Issue's Brief :

A blade is getting a PXE boot connection, but failed with an error :

TFTP cannot open connection

Thanks a lot in advance

[Evan Mickel]

Verify that the appropriate ports are open between the PXE server and the host that you're attempting to boot.  Keep in mind that even if they reside within the same chassis that you may be utilizing the upstream network depending on the pinning of the vNIC.  Inter-fabric connectivity relies on the upstream network.

It would also be helpful to run a packet capture on the PXE server to verify that traffic is reaching it during the boot attempt.

I do see that the host seems to receive an IP address but shows ARP timeouts, perhaps the ARP traffic is not reaching the PXE server or being propagated as it should.

Thanks!

[Ron Baduach]

Hi Evan, thanks for your reply .

A VM is succeeded to PXE boot

and note, that the VM is sitting on a host that have a NIC [on the service-profile] with the same pin group like the physical blade's service profile .

I'll try maybe to run a packet capture later

But, is there something else you can thing about ?

[LukaszTJB]

Hi Ron,

check the configured DHCP lease for the blade server. Are Default Gateway and PXE Server configured correctly? The ARP timeout indicates there is an IP- Address not reachable. I assume it's the default gateway and the PXE Server is located on another sub net.

If you were facing a Network issue on Layer2 you typically won't get an IP-Address from DHCP.

[Ron Baduach]

Hi,

Yes, the DG and PXE Server are on a different subnets

But, we can see on the screenshot that the PXE is succeeded to "talk" to the server

That's what strange here

what/where can some protocol be blocked between the blade until the VM ?...

[Ron Baduach · Accepted solution]

Hi,

After lot of troubleshooting, I decided to re-install the BMA .

And, after the BMA re-installation - it's working !

we can PXE boot a blade !

thank you all