IGMP multicast issues after upgrade 4.3 > 4.4

[DynFi User]

Hi all,

We are having some strange issues with IGMP.
We have a three cluster node running v.4.4 (latest stable)

The Cluster is working and we have managed to have everything running.
Server indicates that : Cluster: pmox-osnet, Quorate: Yes.

Problem is that we have experienced a very high latency on nearly all nodes of our cluster.
This has the following effect : high latency during typing - difficult to impossible typing on the cluster nodes (for some times).


Server are high-end SuperMicro server with 100 to 256GB RAM - so this can't be the cause of our problems.

Even stranger : when we issue

root@proxmaster:/home/xxx# omping -c 10000 -i 0.001 -F -q 192.168.210.10 192.168.210.11 192.168.210.12
192.168.210.10 : waiting for response msg
192.168.210.12 : waiting for response msg
192.168.210.10 : waiting for response msg
192.168.210.12 : waiting for response msg
[…]

So the multicast ping won't get answered ?!
But the cluster works !

We can migrate the VM correctly from one cluster node to another (using NFS Mount sharing points).


Our cluster is configured using OVS.
We have various network (including the network where cluster nodes are installed) which are configured using OVS. Configuration looks like this :


auto lo
iface lo inet loopback

allow-vmbr0 eth0
iface eth0 inet manual
ovs_type OVSPort
ovs_bridge vmbr0

allow-vmbr1 eth1
iface eth1 inet manual
ovs_type OVSPort
ovs_bridge vmbr1

allow-vmbr0 bond0
iface bond0 inet manual
ovs_type OVSBond
ovs_bridge vmbr0
ovs_options bond_mode=active-backup

allow-vmbr1 bond1
iface bond1 inet manual
ovs_type OVSBond
ovs_bridge vmbr1
ovs_options bond_mode=active-backup

auto vmbr0
iface vmbr0 inet manual
ovs_type OVSBridge
ovs_ports eth0 bond0

auto vmbr1
iface vmbr1 inet manual
ovs_type OVSBridge
ovs_ports eth1 bond1 admin

allow-vmbr1 admin
iface admin inet static
address 192.168.210.11
netmask 255.255.255.0
gateway 192.168.210.1
ovs_type OVSIntPort
ovs_bridge vmbr1
ovs_options tag=210



Do you know how we could have the omping working and better understand the lagging problem that we have ?


Thanks for your answer.

 

[mir]

Our cluster is configured using OVS.

An OVS bridge cannot act as an IGMP querier so you need to enable an IGMP querier on your physical switch or router. See https://pve.proxmox.com/wiki/Multicast_notes Especially the part with info on how to configure your switch.

 

[DynFi User]

Ok this is where we were heading toward.

Thanks for your confirmation.
Will post the detailed conf of our HP switch once It is done.

 

[DynFi User]

That being said, I would be interested to know why this seems to happen after the reboot and upgrade of the system from 4.3 to 4.4 ?
What is the root cause of this IGMP problem ?

Because we didn't notice such problem before the update.