It’s never an issue with DNS! Even when DNS looks right…it’s still DNS! I came across an issue today trying to upgrade a 6.5 VCSA to 6.7. The new VCSA appliance deployment was failing with an OVFTool error suggesting that DNS was incorrectly configured.
|
1 2 3 |
OVFTool[30736] [Originator@6876 sub=Default] Curl_perform error code 6 (Couldn't resolve host name) OVFTool[14004] [Originator@6876 sub=Default] Closing image source and targets OVFTool[14004] [Originator@6876 sub=Default] CURL error buffer: Could not resolve: tpm00-104.aperaturelabs.biz (Domain name not found) |
Initially I used the FQDN for source and target vCenter’s and let the installer choose the underlying host to deploy the new VCSA appliance to. Even though everything checked out fine in terms of DNS resolution across all systems I kept on getting the failure. I triple checked name resolution on the machine running the update, both vCenter’s and the target hosts. I even tried using IP addresses for the source and target vCenter but the error remained as it still tried to connect to the vCenter controlled host via it’s FQDN resulting in the error.
After doing a quick Google search and finding nothing, I changed the target to be an ESXi host directly and used it’s IP address over it’s FQDN. This time the OVFTool was able to do it’s thing and deploy the new VCSA appliance.
The one caveat when deploying directly to a host over a vCenter is that you need to have the target PortGroup configured as an ephemeral…but that’s a general rule of bootstrapping a VCSA in any case and it’s the only one that will show up from the drop down list.
While very strange given all DNS checked out as per my testing, the workaround did it’s thing and allowed me to continue with the upgrade. This didn’t find the root cause…however when you need to motor on with anupgrade, a workaround is just as good!
