none
MDT 2013 - A Connection to the deployment share could not be made - DHCP Lease was not obtained RRS feed

  • Question

  • First, let me give you some context:

    Framework: MDT 2013 with MS SDK 7.1

    Task Sequence: Standard Client TS with sysprep and capture.

    Target workstation (build workstation): VM Guest on ESX 5.5 host, 8 vCPU, 8GB RAM, LSI Logic SAS Controller, E1000 NIC, SSD DAS

    Behavior: The VM loads and installs the OS fine in PE, VM boots into OS successfully and resumes the TS, after the first system reboot, the error message occurs and it reads:

    A connection to the deployment share (\\*********\DeploymentShare$) could not be made. DHCP Lease was not obtained for any Networking device! Possible Cause: Check physical connection. Retry:...... Cancel:.....

    While observing this error, I didn't notice the NIC hadn't completely initialized and obtained an IP yet (network adapter icon in systray), additionally hitting retry after the NIC was initialized resumed the TS.

    This behavior reoccurs with several subsequent reboots until a few more applications (Citrix Receiver, VMware Tools) with services are installed which seem to then slow the system boot-up time and then allows the TS to start after the NIC has initialized.

    From several posts I've read on this forum, this particular behavior was alleviated by a "wait for IP lease" mechanism built into the TS engine which was introduced in MDT 2010 SP1, I wasn't able to find any other confirmation whether this mechanism is still in effect with MDT 2013. Another point worth mentioning from several other posts I was able to find is that this behavior appears to manifested itself on target workstations with SSDs, which would somewhat explain the faster TS load time vs waiting for an IP lease. I've also tried to replicate this behavior in a non-SSD and low-performance VM environment and I wasn't able to replicate it.

    My question: Does anyone else have experienced this behavior with MDT 2013 and if so, how did you resolve it? Or is this a bug?

    Monday, October 20, 2014 6:50 PM

Answers

All replies

  • Check out this thread and maybe it will solve your problem!

    https://social.technet.microsoft.com/Forums/en-US/db9f6e36-5cc4-44c0-a87b-fe2c090115db/mdt-2010-update-1-wizard-error-dhcp-lease-was-not-obtained-for-any-network-device?forum=mdt

    Monday, October 20, 2014 8:52 PM
  • This fix doesn't seem to apply to my case. The MDT server pings just fine after the network adapter initializes.
    Monday, October 20, 2014 9:06 PM
  • I have this issue intermittently as well.  For us, it coincided with the deployment of IP phones, which meant PoE switches all around.  However, the problem persisted even after we turned off PoE to the ethernet ports from which we normally PXE boot.

    As this issue has been intermittent, I've backburnered it.  When it does happen, I just wait for the lease to arrive then rerun the wizard.

    Tuesday, October 21, 2014 3:39 PM
  • I have this issue intermittently as well.  For us, it coincided with the deployment of IP phones, which meant PoE switches all around.  However, the problem persisted even after we turned off PoE to the ethernet ports from which we normally PXE boot.

    As this issue has been intermittent, I've backburnered it.  When it does happen, I just wait for the lease to arrive then rerun the wizard.


    Thanks for the feedback, that's true the TS can be resumed manually once the lease has occured but it defeats the purpose of an automated TS if I have to keep an eye on it and intervene if I need to.
    Tuesday, October 21, 2014 6:13 PM
  • Another point worth mentioning from several other posts I was able to find is that this behavior appears to manifested itself on target workstations with SSDs, which would somewhat explain the faster TS load time vs waiting for an IP lease.


    With all due respect, I would not blame the SSD drives, as much as I would question your network.

    Is your DHCP server just slow and/or overwhelmed? What about PortFast?

    Get your network guys to document the latency of DHCP request for your location. If necessary, have them do a netmon trace.


    Keith Garner - Principal Consultant [owner] - http://DeploymentLive.com

    Wednesday, October 22, 2014 7:05 AM
    Moderator
  • I have this issue intermittently as well.  For us, it coincided with the deployment of IP phones, which meant PoE switches all around.  However, the problem persisted even after we turned off PoE to the ethernet ports from which we normally PXE boot.

    As this issue has been intermittent, I've backburnered it.  When it does happen, I just wait for the lease to arrive then rerun the wizard.


    Thanks for the feedback, that's true the TS can be resumed manually once the lease has occured but it defeats the purpose of an automated TS if I have to keep an eye on it and intervene if I need to.
    The network guys here recommended putting wireshark or network monitor on it and figuring out just what the heck is going on.  Basically, what Keith Garner said.  They also disabled PortFast awhile back to see if that made any difference, and it did not.

    • Edited by Atreus21 Wednesday, October 22, 2014 2:26 PM
    Wednesday, October 22, 2014 2:25 PM
  •  They also disabled PortFast awhile back to see if that made any difference, and it did not.

    "Disabled"  ... "awhile back"  hold on...

    did they re-*ENABLE* PortFast? :^)


    Keith Garner - Principal Consultant [owner] - http://DeploymentLive.com

    Friday, October 24, 2014 5:42 AM
    Moderator
  • Try here -

    https://social.technet.microsoft.com/Forums/en-US/5656b654-3f6d-444f-b650-f8dd2ff933d5/mdt-2010-deploy-dhcp-error?forum=mdt

    My fix is proposed and confirmed by others there; sounds like it's relevant to your problems due to you only seeing the issue with SSDs.

    • Proposed as answer by benweston Saturday, November 15, 2014 12:37 AM
    • Marked as answer by Cha Lo Sunday, November 16, 2014 6:04 PM
    Saturday, November 15, 2014 12:36 AM
  • Try here -

    https://social.technet.microsoft.com/Forums/en-US/5656b654-3f6d-444f-b650-f8dd2ff933d5/mdt-2010-deploy-dhcp-error?forum=mdt

    My fix is proposed and confirmed by others there; sounds like it's relevant to your problems due to you only seeing the issue with SSDs.

    That did it, that's actually what I've been trying to figure out, is where to pause the TS from resuming.

    Thanks, that was AWESOME!

    Sunday, November 16, 2014 6:05 PM