2

enter image description hereThis is my storage network:

ESXi Host(Dual 10gb NICs) >> 2xNetgear xs712T >> Synology RS10613xs+ (Dual 10gb NICs)

ESXi vmnic1 >> NG Switch 1 >> Syn NIC 1
ESXi vmnic2 >> NG Switch 2 >> Syn NIC 2


Hello.

My Synology device performs better using NFS instead of iSCSI. A lot better.

The simple text diagram above is how I have my devices connected at the moment with a test ESXi host. My intention is to have this setup in my production environment if I can get it working how I want it.

My requirements: Have an NFS Share mounted as a data store on my hosts survive one of the Netgear switches failing.

The 10gb nic in the ESXi 5.5u2 host is an Intel 82599 10gb TN Dual Port card that is on the VMware HCL.

The Netgear xs712T switches (I know) do support the LAG protocol but not across separate switches.

The Synology RS10613xs+ (DSM 5.1-5022) does support the ability to Bond the two NICs into either a LAG or "Active/Standby".

I'm not sure this is technically possible. Synology support initially said it should work but couldn't give me any examples. Netgear support suggested factory resetting the switches at the first sign of trouble and didn't know anything really (fair enough they're not VMware experts). I don't have VMware support.

So to configure a LAG between the Synology NICs and the switches is pointless as I only have one physical link per switch making a LAG useless. Am I correct?

To create a LAG in this scenario I would need switches that support etherchannel/LAG/LACP across multiple separate switches e.g. top end HP/Cisco

Therefore I could get another dual port 10gb NIC in my host to try two separate LAGs across two separate switches but then I cannot get another 10gb NIC into my Synology :(

So I've setup my Synology with the two NICs bonded as "Active/Standby" which has generated one single NIC interface with a single Virtual IP that I present to my teamed 10gb NIC on my host.

My host NIC is set like this:
1 vSwitch
1 VMKernel with both 10gb Adapters
NIC teaming:
Load Balancing: Use explicit failover order
Failback: No
ALL OTHER SETTINGS DEFAULT

I can ping from my Synology to my ESXi and vice versa. I can remove a Netgear switch and the pings remain working. Yay!!

But when I then mount an NFS store and test the store goes offline and the VMKernel logs show: Device or filesystem XXXXXXXX has entered the All Paths Down Timeout state

I tried lowering the NFS.MaxQueueDepth to 64 which made it slightly less crashy. But still cannot survive a Netgear switch being removed.

Am I trying to do the impossible with what I have?

This is my first post, please go easy. Thanks

  • 1
    You don't need LACP for vSwitch link failover. And in this case, you won't link LACP on the storage side either. Just use standby NICs. – ewwhite Feb 20 '15 at 12:21
  • I tried the configuration with a new vSwitch with default settings, added VMKernel with default settings and made one NIC active and one NIC Standby. Didn't work :( – TSfotec Feb 20 '15 at 14:49
  • Did you try with both NICs active? – ewwhite Feb 20 '15 at 15:11
  • Yes. I tried with both Active and no dice. – TSfotec Feb 20 '15 at 16:56
1

I would do this with standard vSwitches and possibly NIC primary and standby adapters. You don't need LACP for this.

Create a dedicated vSwitch. Make one or both NICs active... And I think that's the most you can do with this setup, since your switches aren't stackable. Let me look into it, though.

enter image description here

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.