As a storage specialist, my skill sets is really limited to basic networking to the level where I needed to get a cluster up and running. Ironically, I started my career a decade ago being a CCNA and I did a fair share of CCNP related training.
Anyway, this is gonna be a quick post for those (who like me, with limited networking skills) troubleshoot NSX Host Preparation.
I came across an error like below when I was trying to push out NSX to the hosts in the cluster. I also checked to ensure that NSX Manager was able to ping and resolve the host names as many posts suggest that it could be related to resolution. All of the above checked out.
esx2.cclab.com: Unable to access agent offline bundle at https://nsx-manager.cclab.com/bin/vdn/vibs-6.4.5/6.7-13168956/vxlan.zip.
Cause : <esxupdate-response>
<errorDesc>Failed to download metadata.</errorDesc>
<msg>('https://vcenter.cclab.com:443/eam/vib?id=76d8ace7-9183-469a-94c4-85d66e829a4f', '/tmp/tmpqsdnxvsa', '<urlopen error [Errno -2] Name or service not known>')</msg>
After troubleshooting for a couple of hours, it occurred to me perhaps the hosts themselves were actually the ones not resolving correctly. I checked the hosts /etc/resolv.conf and realised that it was completely blank! This is bizzare, given that I have full resolution with vSAN, vROPS and a slew of services that were DNS dependent.
Once I fixed the resolv.conf file, clicked “Resolve”, it started installing and I was done!
At the end it was very trivial, but hopefully it helps if you have ran out of ideas.
NOTE : Communications Channels, do take a while to transition from UNKNOWN to UP.