Hello
We have a simple setup with 2x cisco WS-C4948-10GE (san-sw3 and san-sw4) connecting NFS storage (appliance1) and iSCSI storage (appliance2) to 6 vmware esx hosts with 1 link on each switch (see attachment for network diagram). These two switches are connected to each other through a 6x1gbit port-channel and have one interface each for management and vtp data.
This setup has been working fine for a while (6+ months) but a few days ago when we removed a VLAN from both switches on the trunk connecting appliance1 (nfs storage). This VLAN was only used for LDAP authenticating users on a network share no longer used.
When this VLAN was removed we lost connection to all NFS storage on the esx hosts, randomly the hosts connected again but quickly got disconnected. I was able to browse datastores for a while and some times single hosts seemed to work (but awful slow). This reminded me of MTU issues that I have seen before on esx hosts however, we did not touch any MTU settings and MTU seems to be configured properly (on esx hosts, on switches and on appliances). Whats even more weird is that we did not lose iSCSI connections, these seem to have been functioning all the time.
In addition I was able to ping appliance1 from all esx hosts with MTU of 1500. When I tried to ping with MTU 9000 some hosts worked, later on I was able to ping from all hosts with MTU 9000 but connectivity was not restored.
To resolve the issue I shutdown the nfs storage ports (appliance1) and portchannel on switch2 (san-sw4), leaving one link between switches (trunk on GigabitEthernet1/46) still up.
I have attached our switch configuration as it was when the issue was live (ie no shutdown ports). I would very much appreciate any suggestions to what actually happened and/or solutions.
Many thanks
/Daniel
