Hello all,
I am currently struggling with setting up a openfabric network across 3 nodes.
One fabric is always showing not ok and will not get up but it is not always the same one.
So the fabric between 1 and 2 is working and 3 is not getting up.
If I now restart node 2 for example the fabric on 3 gets up but 2 stays down and not ok and so forth. So I always only get the connection working between 2 nodes.
Short description of my current hardware:
3 nodes with each 2 network cards for my VMLan.
2 switches at the top that are connected with each other.
Node1:
card 1 -> switch 1
card 2 -> switch 2
Bond2 - active backup both cards
Bridge0 - bond2 (VLAN vnet and VXLan vnet)
Node2:
Node2 - card 1 -> switch 1
Node2 - card 2 -> switch 2
Bond2 - active backup both cards
Bridge0 - bond2 (VLAN vnet and VXLan vnet
)
Node3:
Node3 - card 1 -> switch 1
Node3 - card 2 -> switch 2
Bond2 - active backup both cards
Bridge0 - bond2 (VLAN vnet and VXLan vnet)
On each node I made an active backup bond and connected that bond to a bridge so i can use that bridge in an sdn VLAN vnet and that is working.
If I now define an IP adress on each bridge I can ping across all nodes and MTU is also working fine with jumbo frames without fragmentation.
A vxlan vnet with each node as a peer is also working perfectly.
Now the issue is, that my cluster will grow much more (7 more nodes are on the way) and to manually only setup the peers in a one line input can fast become a bit confusing...
So I wanted to setup a openfabric to streamline the configuration and make it more easy to follow but that leads to the above mentioned issue...
I tried using the bond, I tried using the bridge and I tried using the cards directly but I always get the same issue or nothing working.
I am currently lost and would really appreciate help.
I am currently struggling with setting up a openfabric network across 3 nodes.
One fabric is always showing not ok and will not get up but it is not always the same one.
So the fabric between 1 and 2 is working and 3 is not getting up.
If I now restart node 2 for example the fabric on 3 gets up but 2 stays down and not ok and so forth. So I always only get the connection working between 2 nodes.
Short description of my current hardware:
3 nodes with each 2 network cards for my VMLan.
2 switches at the top that are connected with each other.
Node1:
card 1 -> switch 1
card 2 -> switch 2
Bond2 - active backup both cards
Bridge0 - bond2 (VLAN vnet and VXLan vnet)
Node2:
Node2 - card 1 -> switch 1
Node2 - card 2 -> switch 2
Bond2 - active backup both cards
Bridge0 - bond2 (VLAN vnet and VXLan vnet
)
Node3:
Node3 - card 1 -> switch 1
Node3 - card 2 -> switch 2
Bond2 - active backup both cards
Bridge0 - bond2 (VLAN vnet and VXLan vnet)
On each node I made an active backup bond and connected that bond to a bridge so i can use that bridge in an sdn VLAN vnet and that is working.
If I now define an IP adress on each bridge I can ping across all nodes and MTU is also working fine with jumbo frames without fragmentation.
A vxlan vnet with each node as a peer is also working perfectly.
Now the issue is, that my cluster will grow much more (7 more nodes are on the way) and to manually only setup the peers in a one line input can fast become a bit confusing...
So I wanted to setup a openfabric to streamline the configuration and make it more easy to follow but that leads to the above mentioned issue...
I tried using the bond, I tried using the bridge and I tried using the cards directly but I always get the same issue or nothing working.
I am currently lost and would really appreciate help.