bergquistcompany
asked on
Cisco UCS through Fiber Interconnects running VMWare with NLB can't ping from any other host
We are setting up NLB on VMWare 5.0 Windows 2008 R2 machines we've kicked up and built a cluster. We can ping the 2 NIC IPs but not the cluster IP except from any VM on the same host. Other hosts do not work.
The Cisco goes through Fiber Interconnects so I'm guessing it's not allowing it to pass through due to some MAC issue, but am at a loss.
The Cisco goes through Fiber Interconnects so I'm guessing it's not allowing it to pass through due to some MAC issue, but am at a loss.
Andrew Hancock (VMware vExpert PRO / EE Fellow/British Beekeeper)
multicast or unicast mode?
ASKER
multicast
I can ping from the same host but not from any other.
I can ping from the same host but not from any other.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Ok but if I have 5 Cisco Blades in one Cisco UCS Chassis and they communicate through a fiber interconnect they aren't going out to a physical switch so where do I need to make this in a vswitch? Also any documentation on the commands. I can do this on 2 stand alone machines without issue, just struggling through the UCS.
it needs to be made on the physical switch, nothing todo with vSwitch.
the commands and documentation are in the VMware links in my post.
the commands and documentation are in the VMware links in my post.
ASKER
Here is my issue though the Cisco UCS goes through Fiber Interconnects that connect to the Ethernet uplinks on my cisco switch Eth 1/2. Can I do these commands against the uplink ports?
all ports need to have static arps defined.
ASKER
ok so I have followed link 1.
I am in link 2 and have done step 3. Neither of which seem to apply to a port
I'm a little uncertain on #4
Would I do mac-address-table static 02-bf-0a-0a-03-95 vlan 1 (in the example it shows ge1/1 ge1/2 ge1/15 ge1/16). Looking at my CH-MC FI for fiber interconnects would I use Eth 1/2 Eth 1/1 Eth 1/2 Eth 1/1 or is there a way to say all?
I am in link 2 and have done step 3. Neither of which seem to apply to a port
I'm a little uncertain on #4
Would I do mac-address-table static 02-bf-0a-0a-03-95 vlan 1 (in the example it shows ge1/1 ge1/2 ge1/15 ge1/16). Looking at my CH-MC FI for fiber interconnects would I use Eth 1/2 Eth 1/1 Eth 1/2 Eth 1/1 or is there a way to say all?
when we have had issues we have created static arps on ALL ports uplinks that you would expect to use this cluster IP address on in your network, yes every physical switch.
yes you would assign it to all if those carry the multicast traffic, there is no All value
yes you would assign it to all if those carry the multicast traffic, there is no All value
ASKER
ok so I have followed link 1.
I am in link 2 and have done step 3. Neither of which seem to apply to a port.
I'm a little uncertain on #4. Would I do:
mac-address-table static 02-bf-0a-0a-03-95 vlan 1 (in the example it shows ge1/1 ge1/2 ge1/15 ge1/16). Looking at my CH-MC FI for fiber interconnects would I use Eth 1/2 Eth 1/1 Eth 1/2 Eth 1/1 or is there a way to say all?
CDP.PNG
I am in link 2 and have done step 3. Neither of which seem to apply to a port.
I'm a little uncertain on #4. Would I do:
mac-address-table static 02-bf-0a-0a-03-95 vlan 1 (in the example it shows ge1/1 ge1/2 ge1/15 ge1/16). Looking at my CH-MC FI for fiber interconnects would I use Eth 1/2 Eth 1/1 Eth 1/2 Eth 1/1 or is there a way to say all?
CDP.PNG
yes, you would use ports ethernet nics from ESXi servers are connected to.