Good Evening- Copying files takes one or more ESX servers down? Yes- Using a UNC path to copy a file from VMa to VMb could purple screen your ESX server. All you need to do is use the E1000 VM NIC type. Synopsis> Don’t use E1000 VNIC on ESX 5.x if you use windows VMS? Apparently so!
This problem is not Lync Specific, but I saw this happen with a verifiable pattern. To reproduce this issue, all that was necessary was to copy a large file between virtual machines that had a common storage device (SAN). All Virtual Machines on multiple hosts failed, and a hard power down was required in some cases to return server to functionality. A ticket was put in to VMware and the results were surprising.
1. The Issue was confirmed as a problem from VMware under KB-2059053
2. The Recommendation from VMware was to use the VMXNET3 virtual adapter and reduce the usage of the E1000 series adapter as much as possible.
3. In addition, disable RSS within the Windows virtual machine. For more information, see the Resolution section of the KB article Poor network performance or high network latency on Windows virtual machines (2008925).
4. This problem may occur on ESX 5.0, 5.1 or 5.5
The Simple operation of copying a file between virtual machines could occur under many circumstances so I would add tags to suggest this could effect any windows implementation on VMware. Please beware of this problem as the issue could cause multiple VMware Hosts to fail, taking down all machines on the Host. We saw this occur on a copy of Lync Update.exe from one server to another. I placed this in the Lync Blog section of my posts, but any Windows application with the VMware virtual NIC E1000- Is susceptible to this problem.
The cause is listed as- This issue occurs when the rx Ring buffer fills up and the max Rx ring is set to more than 2. The next Rx packet received that is handled by the second ring is NULL, causing a processing error.