multicast problem after upgrade from 1.8 to 1.9
- Thread starter lozair
- Start date
yes there is the same problem with 2 vms on the same proxmox host on the same bridge
we don't use openvz but i will attempt to test it....
Just to be sure i'm correct.
I make 2 openvz containers and i connect them to the ethernet bridge ?
Last edited:
Yes, but wait. It would be even better if we can reproduce the problem at the host level (without any VM/container) involved. Do you think that is possible somehow?
Anyways, using openvz container is a good test anyways, because it uses veth instead of tap devices.
Did you already tried using another multicast address (just to make sure there are no address conflicts)?
Did you already tried using another multicast address (just to make sure there are no address conflicts)?
sorry i have not see your response.
I maked the test and it's the same problem.
With 2 openvz containers, one sending packet and the other receiving packet, nothing appear at the bridge level........
Guest A ----> | proxmox bridge | <---- Guest B
Guest A send the packet, tcpdump on bridge get nothing, tcpdump on guest B get nothing ( logic
)To test it whithout vm :
I have an unused network card, i can connect it to the same LAN of my servers and test if i receive the packet at the physical interface level.
My assumption is yes because of we can always capture multicat packet at the bridge level.
That what you think about ?
Guest A send the packet, tcpdump on bridge get nothing, tcpdump on guest B get nothing ( logic
Isn't that behavior different? You always wrote that you receive correct at bridge level, and now you get nothing on the bridge?
yes you are right.
All test we made before, we sent packet from external server (out of proxmox).
So this test case seems to be easier - please try to find out whats wrong here first.
No i always sent multicast packet from out of proxmox machine.
I only get yesterday the "partial oracle application" to reproduce the problem.
Since yesterday i can copy this "application" on any server to make tests........
But i make a mistake, we receive the packet at the bridge level when the two guests are connected on the same bridge......
sorry i'm not looking on the good interface with my tcpdump....
So we can see the following :
PhysicalHostA --> LAN ---> proxmoxBrige --> GuestB : We can capture packet at the proxmoxbridge level but not on the GuestB
GuestA ---> proxmoxBrige --> GuestB : We can capture packet at the proxmoxBridge level but not on guestB
Last edited:
We need to reproduce the bug without external components involved (at least no external switches - direct patch cable connection would be OK).
I confirm the problem arise with 2 vm on the same proxmoxbridge
The question is now what problem??
A.) packets does not arrive at the bridge or
B.) packets does not arrive at the target
Sorry if i'm not clear....
I make the two scema in ordre to clarify the problem :
The first show than when we issued the multicast from an outside host, the proxmox bridge receive the packet but the guests never receive any packets.
The second show than when we issued the multicast from a guest inside proxmox, the bridge receive the packet, but the other guests never receive any packets.
The amazing thing it's in the second schema we can see than the outside host which is on the same LAN receive the packet, and respond with another multicast packet using the same port and same multicast adress.
In this case the bridge receive the 'response' packet and the guest which sending the "initial packet" receive the response !!!!
The other guest never receive any packet...
