[OmniOS-discuss] NFS Datastore vmware esxi failover

[Matt Weiss]

Well thought out, but I have accounted for that.  I have even waited 
over night.

I can run   vmkping 172.16.50.100  while I run my ip setting scripts 
manually and merely drop a ping at most.

I can also mount the datastore as a new one right afterwards and it 
mounts fine.  I tried on another ESXi host and it creates the same UUID 
datastore name so I think I can rule that out as well.

I am wondering if I need any sort of special setting on the sharenfs 
property?  This is all being done with 10GbE dedicated switch with 
nothing else on it currently.

-Matt

On 11/8/2013 10:36 AM, Jim Klimov wrote:
> On 2013-11-08 17:17, Matt Weiss wrote:
>> I am working on a failover script using OmniOS as a NFS server.
>>
>> According to VMware, if I mount and nfs datastore via its IP Address
>> then I should be able to move the IP around and still mount it, however
>> it is not working right.
>
> As a random guess, can this be related to ARP address caching?
> I think that likely the VMWare hosts continue to connect to the
> same IP address for NFS transactions, which automatically uses
> the MAC addresses that they continue to remember as matching the
> storage IP address, until this knowledge times out or is removed
> by forceful actions.
>
> Did you have a chance to either wait for 5 minutes, or flush the
> old address out of VMWare hosts? Perhaps, if your new active NFS
> host would "ping" the VMWare hosts, this would update their ARP
> entries (unless the hosts or your network have anti-spoofing)...
>
> Alternately, you can try setting the same MAC address on the two
> VNICs of the two hosts, but then you'd have to similarly refresh
> the Ethernet switches' knowledge of address-to-port bindings.
>
> HTH,
> //Jim Klimov
>
> _______________________________________________
> OmniOS-discuss mailing list
> OmniOS-discuss at lists.omniti.com
> http://lists.omniti.com/mailman/listinfo/omnios-discuss