locked
Hyper-V Manager and Hyper-V Virtual Machine Connection Excruciatingly Slow RRS feed

  • Question

  • I have a scenario in which Hyper-V Manager and Hyper-V Virtual Machine Connection are running excruciatingly slow.

    The problem is puzzling for a few reasons.

    • This is all on a LAN.
    • A Remote Desktop Connection from the exact same workstation to the exact same server has absolutely zero latency issues (of course, this means that the Remote Desktop Service has to be exposed and accessible, but that can't be left that way).
    • Similarly, enabling ICMP for "domain" and pinging does not appear to indicate any connectivity or latency issues to either the server guest or the server host.

    The basic environment is as follows:

    • Server Host running Windows Server 2012 R2 with Hyper-V
    • Server Guest running Windows Server 2012 R2 and setup as a primary domain controller
    • Workstation running Windows 8.1 Professional (also has Hyper-V enabled)

    When on the Windows 8.1 Professional workstation and opening the Hyper-V manager, there is a several second lag in doing just about anything related to the remote Hyper-V server host.  If launching a Hyper-V Virtual Machine Connection from Hyper-V Manager, it can take a minute or two for the connection "screen resolution" dialog to appear.  If it makes it past that stage, for the Hyper-V Virtual Machine Connection to respond to just one mouse click on the logon screen (when first connecting), it can take 30 - 120 seconds to respond, if it even responds at all.  Sometimes the "lost connection / retrying" notification even appears.

    Perhaps I haven't quite found the right combination of search keywords, but a large majority of the relevant search results deal with disabling Virtual Machine Queue capabilities; however, that appeared to be an issue primarily related to Broadcom Ethernet adapters, but all the Ethernet adapters on the Hyper-V host are Intel.

    At first glance, given that all other network connectivity appears to be fine, this might seem to be something with the Hyper-V configuration between the server host and the workstation on which Hyper-V Manager is running, but I am at a loss for what could possibly be the culprit.

    Any suggestions?

    Thank you.

    Wednesday, May 20, 2015 2:12 PM

Answers

  • Thanks to this post on ServerFault, which in turn referenced a TechNet forum post, I have discovered that this problem was related to Large Send Offload being enabled on the Hyper-V Virtual Ethernet Adapter (which it apparently is by default).

    Go into the Properties dialog for the Hyper-V Virtual Ethernet Adapter, switch to the Advanced tab, and change "Large Send Offload Version 2 (IPv4)" to disabled.  According to the other posts, "Large Send Offload (IPv4)" should also be disabled, but that property does not exist in the properties list that I see.

    After making this single change, the performance instantly improved to the expected level.  Additionally, Wireshark had been flagging a significant number of out-of-order (and related) frames (aka the black rows), and those have all cleared up.

    • Marked as answer by mesheets Friday, May 22, 2015 12:43 PM
    Friday, May 22, 2015 12:43 PM