Jump to content
Welcome to our new Citrix community!
  • 1

Strange PXE Issue on Citrix Hypervisor 8.1


Mark Federico

Question

So we are having a stange issue with PXE booting on a new Citrix Hypervisor Server Stack I am setting up

 

Sometimes VMs will PXE other times they wont one thing that always works is if you create a New empty VM (VM 1) and its first boot is on Citrix Hypervisor Host (Host 1) it will always PXE boot that first time but if then you go to boot that VM 1 on Host 2 it may or may not PXE boot (it boots about 25% of the time) if you then go back to Host 1 it will PXE boot fine but the other weird thing is if you reboot Host 1 that same VM that always booted on Host 1 will now have the same issue on Host 1.

 

We know its not a DHCP or Switch issue because we have the same config on our current XenServer 7.1 Hardware Stack and all those VMs PXE boot fine on any host in that stack using the same DHCP servers we are trying to use on this new Hardware stack and the backend vswitch config is also the same. 

 

Another interesting thing is that if we create a VM on our management network which is on the same network as the DHCP server itself that VM will pull an IP on that VLAN from PXE 100% of the time on any host so its only VMs that are on a separate VLAN then the DHCP server

 

We also took wireshark traces n those traces when it works we see the VM requesting an P from DHCP, then DHCP offers the address, the VM takes it and boots. When it doesnt work we boot the VM, The VM then requests an IP from DHCP, However the VM never takes the IP so you see DHCP retry a couple time saying here is the offered IP but for some reason the guest VM never accepts it so we can see it getting to DHCP and the IP offer coming back to the VM but we cant figure out why the VM won't take the offered up IP.

 

We have a case open but the engineer I am working with is equally confused and verified we have everything configured correctly and looked at the wireshark and agreed it was very odd behavior hoping someone might have seen something similar since we want to get this new Hardware deployed but cant do that right now.   

 

Another interesting thing to add to the puzzle working with my Network team today seems like the Citrix Hypervisor host isn't consistently updating the CAM table on my swtich meaning that when it fails and we see my DHCP server send back the IP to the guest if its on a different host its sending it back to the wrong port

 

Example VM was first booted on Host 1 and booted fine. Shut VM Down Booted VM on Host 2 we see the VM reaching out to DHCP and DHCP replaying back and offering an address but it is doing that on the Host 1 port because the CAM table on the switch was never updated.

 

Link to comment

9 answers to this question

Recommended Posts

  • 0

Mark, we have the same issue here on a 8.1 update from 8.0.  After that update, PVS accelerator didn't work until I did another reboot of the host for some firmware updates a week later.

 

Did you find a solution for this?

 

My tcpdump shows the same thing as yours and the odd thing is that once it drops to the UEFI Shell, DHCP works like it should.

 

I've opened ticket 79496886 for this and one other weird PXE boot issue we have where some of our BIOS VMs will now no longer boot but some will.   All of these VMs are on the same VLAN / Subnet.

 

 

Link to comment
  • 0
7 minutes ago, Steven Brown1709154220 said:

Mark, we have the same issue here on a 8.1 update from 8.0.  After that update, PVS accelerator didn't work until I did another reboot of the host for some firmware updates a week later.

 

Did you find a solution for this?

 

My tcpdump shows the same thing as yours and the odd thing is that once it drops to the UEFI Shell, DHCP works like it should.

 

I've opened ticket 79496886 for this and one other weird PXE boot issue we have where some of our BIOS VMs will now no longer boot but some will.   All of these VMs are on the same VLAN / Subnet.

 

 

No still have a case open We did more digging and if you look at it from a switch perspective the issue is the Host isnt updating the switches CAM table so DHCP is replying but because the CAM table isn't updating the DHCP offer is actually being sent to the wrong host which is why the VM never picks it up.

 

As a work around that has seemed to work well we have moved our PVS VMs to boot via BDM (which also uses PXE but a seemly different flavor) but using BDM boot we have seen no issues at all so we basically redeployed all our VMs to boot Via BDM instead of PXE which has been a positive anyhow and slightly improved boot times.

 

We do still have the ticket open (79481886) but even once solved we will probably stick with BDM   

Link to comment
  • 0

I have had the same experience thus far the tech I was talking with didn't really understand the issue. I already am on the second tech the first said Citrix doesn't make PXE so they couldn't help and that I was also running in a non supported configuration needless to say I was also not encouraged.

 

That's one of the reasons we were happy BDM is working well as with some of these Citrix cases where you need to get deep into it they take forever to talk with someone who can actually help.  

Link to comment
  • 0

Mark, we've moved to using BDM partitions the same as you.    Do you have a 10 second delay between when it says it's connecting to the server and it actually doing it?

 

I can see a DNS request from the target device and a response from the DNS server with the IP addresses and then it sits there for 9-10 seconds and does nothing.   

 

After the 10 seconds are past, we can see UDP packets to the login port on the PVS server and it continues on normally.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...