how to set up esxi to do failover in a switchless setup ?

Hi,


My v7 cluster seems to work fine, but I can't figure out how to set up vsphere esxi 5 so that the datastore remains available when a resource pool moves to the other san.

I'm using a switchless setup: got a dual port 10GE nic in the ESXi for my storage network, 1 direct connection to each SAN.
When I do static discovery with the virtual ip (10.0.50.100) it finds the datastore.
But when failover happens that path is no longer available. If I then rescan the HBA it will discover the new path, but so it does not happen automatically.

Esxi doesn't allow me to do NIC teaming so I had to created 2 vmkernel ports (iscsi-A and iscsi-B), each bound to 1 physical nic.

Once it has learned both paths it seems to work, but this is not ideal since I have to trigger a failover just to do the setup and I don't think it will stick after a reboot of the esxi.

Am I missing something ?

You can visit our website at: http://www.open-e.com/service-and-su...ts-and-videos/
And look for :
- HA Solution with VMware and Open-E DSS V7 as Virtual Storage Appliance
- How to install Open-E DSS V7 as virtual machine on VMware

You also can find other useful videos there that should help you.

Is this happened only on the secondary system you have? is your ping node was running well ? also can you be sure that this static IP is not been used with any other network devices ?

Please make sure beacon probing is enabled in ESX:
http://kb.vmware.com/selfservice/mic...rnalId=1005577

Hi,

I was unable to make it work, I started from scratch but still the same issue.
I'm using vsphere 5 (esxi 5) and it only allows to have 1 active nic (the other one must be set to "unused") per vmkernel port.
So I guess the beacon probing/nic teaming failover is not kicking in.

See http://blogs.vmware.com/vsphere/2011...i-binding.html

The setup in the video mentioned above is also slightly different in the network setup.

I am however attempting this blog post from Janusz http://blog.open-e.com/is-it-possibl...lable-cluster/
But unfortunately it doesn't show the whole setup and it has probably changed since esxi 5.0.

Any more ideas ?

Hi finally solved the issue.
I figured out that I didn't need iscsi port binding after all, that's only useful for MPIO and since I'm using a single 10 gbit path I don't need that.
So I could leave the nic's in active-active with beacon probing and everything works !

Thanks for the pointers !