No_Way
Enthusiast
Enthusiast

vCloud Appliance interfaces not working

Hi all,

I have a stupid and specific issue in one of my environments and just like to know if any have seen this.

As information, we have around 5 vCD in our environments.

Install a new vCloud Director 10.2.2 (also tested with 10.2) and I can reach eth0 if eth1 is up. If I disable the network in the eth1, eth0 is reachable. If I enable eth1, then IP on the eth0 is not reachable and only eht1 is reachable.

In the beginning, I was using different subnets, but then I try in the same subnet, and I still have the same issue.

This is really strange behavior and I cant see where is the problem.

I thought this could be a bug in the latest vCD, but I try 10.2, same issue.

Mainly I cant have both network interfaces in the appliance enable.

I dont have this problem in any of the other vCD(not running on the same vCenter/ESXi hosts).

0 Kudos
7 Replies
Sreec
VMware Employee
VMware Employee

Eth0 and Eth1 must be in different subnets.  Most likely in your case, Eth0 is configured with a gateway. So routed traffic will egress via gateway IP of Eth0. By default, ETH1 is used for database replication. This is the reason you are not able to access or getting packet drops. Why do you want to reach Eth1? 

Cheers,
Sree | CKA|CKAD|VCIX-3X| VCAP-4X| VExpert 5x
Please KUDO helpful posts and mark the thread as solved if answered
0 Kudos
No_Way
Enthusiast
Enthusiast


@Sreec wrote:

Eth0 and Eth1 must be in different subnets.  Most likely in your case, Eth0 is configured with a gateway. So routed traffic will egress via gateway IP of Eth0. By default, ETH1 is used for database replication. This is the reason you are not able to access or getting packet drops. Why do you want to reach Eth1? 


Hi Screec,

Thank you for your reply.

Yes I know how eth0 and eth1 works.

Like I said, if eth0 and eth1 are enable, eth0 stop working. Only eth1 is reachatble, if I disable eth1, then eth0 is reachatble. And yes yes has a Glocal gateway.

Also I know that we can change IPs, gateways etc, using vami sripts.

And I have move vCDs and even with differnt subnets, both are reatachable. But is not my concern here, but the issue that if both network interfaces are enable I loose connection to eth0.

I only have this issue, in this vCenter with this ESXi hosts. That is why I am trying to understand where is the root cause of this (physical? virtual networks?, vCD networks?)

I have try deploy 10.2, 10.0, even a 9.7 I always get this issue.

Is a very strange behavior.

0 Kudos
Sreec
VMware Employee
VMware Employee

1. Like I said, if eth0 and eth1 are enabled, eth0 stop working.   only eth1 is reachable,

Is this the case from the beginning?  If your appliance deployed is done successfully both the interfaces will be UP and you will be able to reach Eth0. 

How are you testing the connectivity?   During the testing source and destination is on a different subnet or the same subnet? 

Can you take a direct console to VM and check if you can reach any other IP in the same subnet? 

Can you take a direct console to VM and check if you can reach the gateway IP?  

 

 

Cheers,
Sree | CKA|CKAD|VCIX-3X| VCAP-4X| VExpert 5x
Please KUDO helpful posts and mark the thread as solved if answered
0 Kudos
No_Way
Enthusiast
Enthusiast


@Sreec wrote:

1. Like I said, if eth0 and eth1 are enabled, eth0 stop working.   only eth1 is reachable,

Is this the case from the beginning?  If your appliance deployed is done successfully both the interfaces will be UP and you will be able to reach Eth0. 

How are you testing the connectivity?   During the testing source and destination is on a different subnet or the same subnet? 

Can you take a direct console to VM and check if you can reach any other IP in the same subnet? 

Can you take a direct console to VM and check if you can reach the gateway IP?  

 

 


-- How are you testing the connectivity?   During the testing source and destination is on a different subnet or the same subnet? 

Yes, inside VM console using ping -I

-- Can you take a direct console to VM and check if you can reach any other IP in the same subnet? 

Yes all test are done inside the VM console

-- Can you take a direct console to VM and check if you can reach the gateway IP?  

Yes, I can, only if using eth0, since eth1 is a different subnet.

But just a clarification, using different subnets for eth0 or eth1 is only "mandatory" if you are using DB high availablity. If using as a single standalone vCD no need to do that. As you said, is not in use.

0 Kudos
Sreec
VMware Employee
VMware Employee

1. DB sync is L2 communication  and that is why we don't need a gateway for Eth1 

2. If you need L3 connectivity to Eth1 - we need to add static routes ( you will see this option during deployment) or after the deployment via CLI. 

3. Older versions of VCD  support the same subnet for both the adapters which is not the case anymore with Photon OS due to strict reverse path filtering 

4. If you are not able to reach Eth0 IP from any routed subnet, this is a routing or VLAN configuration issue 

5. If you are not able to reach Eth1 IP from any routed subnet, this is because there are no static routes in the OS, and traffic gets dropped 

Cheers,
Sree | CKA|CKAD|VCIX-3X| VCAP-4X| VExpert 5x
Please KUDO helpful posts and mark the thread as solved if answered
0 Kudos
No_Way
Enthusiast
Enthusiast

If is it a routing issue? Maybe, but still don't know here.

But the main problem here is if both are enabled then there is no eth0 network. If I disable eth1 then there are no issues with eth0 routing. This is the main issue here.

PS: Same VLAN / Subnet is used for vCenter, ESXi hosts. Same network, same Virtual Switches/Ports no issues with routing.

0 Kudos
tititototata
Contributor
Contributor

hello, 

I have exactly the same problem with 10.3 version

etho 0 and 1 is different network and different VLAN, 

Eth1 is rechable not eth0 ...

have you a solution ? 

thank's in advance

0 Kudos