After our recent upgrade of vCloud Director from 1.5 to 5.1.2 we started to see some errors during our automated provisioning of VSEs that included the sub allocation of IPs…When creating a new VSE you will see an error similar to what shown below.
Basically the bug incorrectly reports available IPs as being in use. Our devs have put in a workaround that blackholed those IPs so they are not used, but it’s very much a reactive workaround.
In the release notes for vCloud 5.1.3 we found the following fix listed
- Unused external IP addresses from sub-allocated IP pools of the gateway failed after upgrading from vCloud Director 1.5.1 to vCloud Director 5.1.2
After upgrading vCloud Director from version 1.5.1 to version 5.1.2, attempting to remove unused external IP addresses from sub-allocated IP pools of a gateway failed saying that IPs are in use. This issue is resolved in vCloud Director 5.1.3.
Brilliant right! So we deployed in the lab and found that we still came across the error. After getting a VMware SR going and uploading the CELL Logs Global Support came back today and said that this was actually not resolved in 5.1.3 and the bug remains in play. The lead vCloud Dev indicated that this was released in error and the notes will be updated to reflect the fact a fix is not contained in the 5.1.3 release.
No ETA on the fix.
Over the past couple of months we have seen a number of support requests coming into our HelpDesk around network connection dropouts in Windows Server 2012 Guest systems. Common factor in all these requests where E1000 NICS.
There are known issues with the E1000 NIC and Windows 2012 on ESXi 5.1 and above and the solution to the issue is to use a VMXNET3 NIC over the E1000…and before the VMware bashers start to take aim, lets not forget that the E1000 drivers are written by Intel and are provided as such. The VMXNET3 NICs are VMware written and also provide additional advantages such as better CPU efficiency and the ability to use Jumbo Frames.
Because vCloud Director’s default NIC Type is the E1000, if users are not switched on, then Guest VMs running 2012 will potentially have these issues. To get our support guys proactively advising clients of this issue I put together a quick @CloudPhysics Card using the Card Builder that will report on the Operating System Type, list the Guest NICs and their Type.
Which gives you a report as shown below:
The card should be available in the CloudPhysics Card Store shortly, but it’s easy enough to put together as per the builder blocks and filters above.
In a nutshell for Windows Server 2012…use the VMXNET3! There is really no excuse not to…when provisioning VMs in vCloud Director, be aware that the default NIC is the E1000…Maybe a feature request for the next version vCloud Prod Team? As an additional step, vCloud Users should look to template where possible and include the VMXNET3 NIC as part of the build.
For more info there is good guidance in this KB on selecting NICs based on Virtual Machine Operating System Type…