I have encountered the following issue within a customer environment, where vMotion of VMs was totally not working; stuck at 22% and took a very long time to get there and fail once stalled.
So I looked for the usual curkprit such as :
- Make sure the vMotion VMK interfaces existed ;
- Make sure the vMotion VMK interfaces had their static IP properly configured ;
- Make sure the hosts could reach each others through the vMotion network stack.
Following a complete success of the above, I noticed the issue was only affecting a few hosts within the cluster.
I went further looking at the switch ports configuration, and I noticed that the affected hosts interface MTU were not set to jumbo frame (mtu 9000+), where all other hosts were set to 9216.
Setting the interfaces MTU of the affected hosts to jumbo frame resolved the issue.
NOTE : It is always a good idea to look at the vSwitch MTU as well, it was set properly in my case but could be another location where a mismatch could occur.