"Unable to find image on Parallels NetBoot servers"

Z

Zak

Bit poster
I am able to Netboot to the created Parallels Netboot image. I am able to walk through the task sequence wizard, and at the step Apply OS X Image it almost immediately fails with the above message.

I have went through the Network considerations KB, and the KB concerning this specific error yet I'm unable to get this working. There is a DHCP Relay agent on my lab switch pointing to the Netboot server. The Netboot server and SCCM Primary are separate servers.
Are there additional configurations that would cause this to fail?

From what I can tell the client is not even making the request to the server for the image. I'm running Wireshark on the Netboot server and not seeing any data coming in after the TS actually starts. I can plainly see the data coming through for the Netboot image. (Obviously or the PMM image wouldn't load)

Any guidance on this would be greatly appreciated!

PS. I've attached logs from the client and Netboot server. If anything else is needed, please let me know.
 

Attachments

Zak,
Please, run Wireshark with 'bootp' filter in the same network there Mac sits. You will see request
pmm-client-to-server.png
Note that PMM uses Vendor Class Identifier - PRLPMM, which is different from Apple's Vendor Class Identifier - AAPLBSDPC. If you limit you DHCP Relay to only Apple's packets then this may be the issue. You have to configure your network so that this packets can reach the NetBoot server.
 
On the last 'inform' DHCP packet that I see coming from the client is this: (attached is the wireshark capture of all bootp packets.)[remove extension .txt to open]
Vendor class identifier: AAPLBSDPC/i386/MacBookPro11,2
I then searched the entire capture from the Netboot server (resides on different subnet) and saw no mention of PRLPMM . Do Cisco switches block some Vendor Classes? We just have the relay agent set as follows:
ip dhcp relay address <ip of Netboot Server>
ip dhcp relay address <ip of DHCP Server>

Can someone provide the additional configuration needed? I can pass it on to the Net Ops team to test it out. I assumed that when you set the dhcp relay address on Cisco switches, it blindly passed ALL DHCP requests to the address you specify.

Please advise. Thanks in advance.
 

Attachments

After further digging I have noticed that the client sends the correct DHCP Inform packet, but it never makes it to the server. I'm consulting with our Net Ops team to figure out if they're being dropped. Has anyone else experienced this? He informed me that DHCP snooping is turned on. These are Cisco switches, but I don't have enough experience with DHCP snooping to know if this will block certain IP Helper requests.
 
May your NetOps team look into router logs? Maybe you'll find DHCP snooping violation records.
Another option: temporary disable DHCP snooping on this router to test is this the reason.

When you booted into NetBoot image, open Startup Disk from Apple menu. Do you see your boot image here?
 
I think DHCP snooping may be the issue.

To answer your second question: Yes I can boot to the netboot image. I can get all the way through the Parallels Task Sequence Wizard up until the Apply OS X Image step starts. From there I can see the DHCP requests leaving the client, but they never reach the Netboot server.
 
Was this issue ever resolved? Seeing the same thing in my environment, just don't want to nag the Network Engineers until I have more info. Was DHCP snooping the issue?
 
My Network engineers could never confirm that it was DHCP snooping. Parallels support would not give me any more information either. They (support) ended up telling me to rule out my network gear, that I would have to place the client on the same VLAN as my server. This is just not possible right now. It will take more time to plumb the VLANs than it is worth.

Parallels support told me there was nothing else they could do until I tried this method. I hope if you can get your NetOps guys involved that you can provide some feedback!
 
Hi Zak,

Have you been able to understand the cause of this error? I'm having the exact same problem. In the pmm_tswizard.log I see this:

02-10 03:30:54.580 W /TsProcessors:662:50f/ Check failed 'Searching OS X image file': Unable to find image on Parallels NetBoot servers

I've raised a ticket with Support who want a wireshark capture and immediately suspect the network. I had a previous Support case where Support also suspected the network and in the end the issue was an unknown bug when capturing a netboot image, but we spent 3 weeks looking at the network and eventually proving the network was fine so I don't believe the network configuration is the issue here.
 
We captured the image just fine, but we are unable to deploy it. We have not been able to rule out the network, but we don't suspect it is a network issue either. Unfortunately, support refused to help us out any further because we could not rule it completely out. (No time right now)
 
I did a Wireshark capture and we are seeing the DHCP Inform message with Vendor Class Identifier PRLPMM but then nothing - no more traffic is captured and the task sequence fails (see image). If I make any progress on this, I'll let you know.
Task_Sequence_Wire_Shark.jpg
 
I'm seeing the exact same thing. Support told me to put our client on the same VLAN as the server, which is not possible at this time. They have KB on it, but it doesn't seem to apply to us.

Please keep us posted what you find!
 
Any more info on this? I'm having the same issure and have yet to be able to deploy using the Task Sequence functionality.
 
Guys, we have been having the same issue. But after much troubleshooting, it turned out that the our DHCP relay is not forwarding subnet broadcasts (x.x.x.255) and from ports other than 68. During this stage, Netboot tries to send out subnet broadcast instead of global broadcast (255.255.255.255) and also from a random client port instead of 68. Working with Parallels,they were able to provide a private build of NetBoot image creator which does a global broadcast for NetRestore. This, along with setting the client UDP port to 68 using the task sequence variable option fixed the issue for us. Parallels is possibly working on an update with these additions/changes, you'd have to check with them to see if this applies to your situation.
 
Just thought this might help others. Thanks to VinuT for all of the troubleshooting . I've just installed
Parallels Mac Management v4.5 Hotfix 2 for Microsoft SCCM build PMA2012-4.5.4.5 (August 19, 2016)
I had to re-create the boot image using the pmm_osd_image_builder from 4.5.4.5 and the task sequence image deployment has now worked.
 
You must log in or register to reply here.
Back
Top