Usually this means the host (or one of them if this is a cluster type setup) doesn't have a connection to one of the controllers.

yes I have 2 esxi hosts. one host have 2 paths to each lun. but one host only have one path due to a failed nic. but why woul this cause the md to send these alerts? one host has 2 paths

It is not the NIC that causes this. The dead NIC causes the array to have to fail the virtual disk over to the controller that that specific server does have a connection to and this is what brings up the message.

Just replace the bad NIC(port) and verify it reconnects (at least 1 connection to each raid controller) and the errors should stop (after 1 more time of moving the virtual disk back to the preferred raid controller).

i now have all paths available to the md, but after moving the virtual disk over to the preferred path, it still moves back. whats going on?