(edit: writing this while I'm still troubleshooting, for progress see other posts)
short version: I keep losing connection to NFS datastores on a Win 2012R2 server, an issue that temporarily fixes itself each time I connect to NFS shares on that NFS server via a Win7 client (??)
Hi, vSphere 5.5 newb here. Running evaluation version with vCenter server and 2 concurrent ESXi hosts (+a DC + a 2012R2 for iSCSI and NFS access) on Workstation 9, and I'm having trouble consistently creating NFS datastores/keeping them connected.
This is in fact the 3rd time that I start typing a message. Yesterday I had trouble creating the datastores on the two initial ESXi hosts that I have created for self-study. Then after fiddling it suddenly WORKED on both of them, while I had the strong impression that it was NOT due to a specific configuration change. I was typing my first post here at that moment, so I discarded it. Minutes later I lost connection to the datastores and from then onwards have been unable to recreate them.
This is the error that I get:
Call "HostDatastoreSystem.CreateNasDatastore" for object "datastoreSystem-192" on vCenter Server "2008vCenter1.corpa.armaged.cu.cc" failed.
Operation failed, diagnostics report: Cannot open volume: /vmfs/volumes/f52d180f-4bc8d903
Because I had been doing a lot of stuff with those hosts, possibly including settings that broke something, I decided it might be a good moment to start over. I reverted the 2012R2 server to a snapshot just before installing iSCSI and NFS services and set that up again. I installed an ESXi3 and ESXi4 from scratch, 8 NICs each, planning for redundant connections to management network, VM network, iSCSI storage network and 2 extra NICs for stuff that I still have to study like vMotion and FT etc.
I try to take care of setting up the networking correctly, using only 4 NIC at the moment:
vSwitch0 with dual uplink, "VM Network" standard portgroup and VMkernel port for management with IP in 192.168.90/24 subnet. (subnet with only the ESXi hosts and one NIC on the vCenter Server)
vSwitch1 with dual uplink with ProductionPortGroup for VMs, and the required "NFS" VMkernel port with IP in 172.22.0/23 subnet (subnet with the clients, 2012R2 server for NFS, and one NIC on the vCenter Server)
I'm using a 2012R2 server for the NFS. Share settings as recommended (anonymous access, R/W and root access...).
Now, before I started typing this mail I had exactly the same problems with ESXi3: I try to connect to an NFS share "2012R2SAN-NFS1" and it fails with the error. The ONLY thing that I do after this failure is that I start up a Win7 VM to verify NFS client connectivity to the 2012R2 server, and I also create a second NFS share "NFSTEST" on the same volume on the server. I do an NFS connection test with the W7 client, successfull. I go back to vCenter server and try again to set up the NFS datastore for ESXi3 with settings(same as the ones I used before when it didn't work, but now initially to the new "NFSTEST" share):
server: 172.22.0.52 (ip address of the 2012R2 server for NFS)
folder: NFSTEST
Datastore name NFSTEST
For whatever strange reason, it suddenly WORKS???? When I try after this with the original 2012R2SAN-NFS1 share, it ALSO works without problems.
UPDATE: I disconnect the datastores, try to connect them again, BAM: no longer works for either!!!
PROGRESS UPDATE2: it consistently turns out that AFTER I do a Win7 NFS connectivity test to the 2012R2 server, I succeed in creating the NFS datastores on the ESXi !!! When I unmount them, I'm not able to recreate them without first doing another NFS connectivity test with the Win7 client?? WTF??
Now, I enabled all NFS eventlogging on the 2012R2, and when it doesn't work I can actually see a successfull mount and immediately afterwards a successfull UNmount event. BTW, yesterday I tried with NFS on a 2008R2 server and I had the same problems there.
Well, at least one step further. 
Anyone suggestions about how to further troubleshoot this from the ESXi side?