I was setting up my home test lab to try out vCenter 6.7 and because I was using different intel processors on my 3 computers I turned on EVC for my cluster and started adding my 3 ESXi hosts. When I got to my 3rd ESXi host it refused to add it as it had my vCenter Server (vCSA) VM running on it and it was running on a more advanced Intel processor that the other 2 servers. This made things more complex as if I shut it down I wouldn't be able to move the 3rd host into the EVC enabled cluster.
So, I decided to shutdown the vCSA and to copy (SCP) the files off the 3rd ESXi server so that I could upload them to the 1st ESXi server that was already part or the EVC cluster and then proceed to power it on using the EVC settings.
The problem I encountered was that an SCP copy was going at 1MB/s and was going to take days to finish. After investigating this it looks like an SCP copy goes through some kind of console and is unusable for large file copies. So, I created a temporary NFS server, added it as a datastore to all of the hosts and then did a Storage vMotion to it, it took under 5 minutes. I then disconnected the vCSA from the 3rd ESXi server and connected it to the 1st ESX server and powered it on in my EVC cluster. I could then also add the 3rd ESXi server into the EVC cluster.
I've only got 1 NIC in my ESXi servers so I know it is not a NIC issue and searching on the internet it looks like loads of people have encountered this unusably slow SCP copy issue.
Out of interest, could I have done this in a more pain free way? I did also try to downloading the files through the datastore browser, but whenever I selected a vmdk file it just kept downloading a small associated text file instead of the large disk file.
Is there a way to fix this SCP slow copy issue? I searched the internet and found no solutions. For people trying out the product it gives a really bad first impression.
Ive had the same experience. A moderator just locked me out of another thread instead of commenting on how to fix it.