I have several ESX 3.5 servers on a HP C7000 Blade chasis and I am having problems with getting one of my VM guest to communicate on a different VLan than our production Vlan's.
I have setup one out of the 8 switch ports on the HP blade chasis to connect to a trunk port that communicatates with both our production VLan's and VLan75 which is our vpn VLan using dot1q encapsulation.
I setup a port group to talk on VLan75 and assigned it to one of my guest, but I am not able to ping any servers on VLan75 nor can any servers on VLan75 ping the guest.
Any ideas or suggestions would be helpful. I hope I have explained this scenario clearly, please feel free to ask any question to clearify my situation.
Make sure that vlan 75 is not the native vlan on the trunk port, the tags for the native vlan are stripped off so they don't route.
You should be able to verify if the VLAN75 is the native VLAN by removing the VLAN id from the portgroup, thereby removing the VLAN tag. If communication is established, then you know that VLAN is native, and you can have it removed from native or not.
That is not completely correct. You can still have a native VLAN for a trunk. It just means that VLAN is not tagged as are the other allowed VLAN's in the trunk list. This may be why you are having an issue, since the native is not tagged, and you are expecting a tagged packet, you see nothing. As I mentioned, removing your VLAN tag for that port group should be a good test, since you won't expect a tagged packet on that interface.
The Vswitch has 56 ports, and the ESX server has access to 2 physical adapters. The HP Blade chasis has a 8 port switch that all of the servers share. I only have one of those ports connected to a trunked port on our Cisco 6509. Are you saying I may need to trunk all 8 ports?
I removed the VLan tag from the port group, and I still am not getting any ICMP traffic. Just so you know there is no firewall turned on the servers that I am pinging nor is the one on the guest.
No, just the 2 ports that those adapters associated to the vswith are plugged into. I had a similar problem recently and that was our issue.
- or to cheque quickly, you could just remove the one adapter from the vswitch that is not trunked as see if a ping works
Then you will definitely need to verify that the VLAN is allowed on that trunk with your network folks. You can try to check the network hint, and see if your required VLAN shows up there. It isn't always there if the pNIC hasn't seen traffic on that VLAN, but it does every now and then offer good information. It will be on the pNIC page, or from the service console, use: esxcfg-info -n |grep -E -i 'name|hint'
If the chassis switches are Cisco 3020's - then you must ensure that the internal ports that the esx blade connects to also has the correct vlan trunked to them. It sounds like your only trunking to one of the external uplink ports.
Similary for virtual connect, u need to ensure that the correct ethernet port in the VC profile for the blade has the correct external network associated with it.
If you are using the HP switches then, although i have never used them, im thinking it will probably be similar - there are internal ports (which the blades connect to) and external ports (which you use to uplink to external switches with) - you need to ensure you dot1q them both correctly.
I am pretty sure you can set them up indavidually, just make sure the physical nics bound to the vswitch see the same vlans. When in VC you should be able to see the vlans associated with each adapter when looking at the physical adapters.