[pve-devel] corosync, multicast problem because of vmbr multicast_snooping enabled

[Alexandre DERUMIER]

>>Again I only presume. Given only a psychical nic I guess multicast is 
>>by default carried unless explicitly stated otherwise. 

I'm reading the cisco and hp documentations, if igmp snooping is enabled, it require igmp queries to work.
So switches need to send igmp queries (it's the default on cisco switch if multicast is enabled)

(igmp snooping need igmp queries).

But having igmp_snooping disable on a physical switch (or linux bridge), should forward all multicast by default to all ports.


Igmp snooping in linux bridge has been added in kernel 2.6.34, and I think multicast worked before that on linux bridges ;)


Now it seem that igmp queries coming from linux bridge can conflict with igmp queries from physicals switchs.

That's why igmp queries are now disable in newer kernels
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=c5c23260594c5701af66ef754916775ba6a46bbc
"Sending general queries was implemented as an optimisation to speed
up convergence on start-up.  In order to prevent interference with
multicast routers a zero source address has to be used.

Unfortunately these packets appear to cause some multicast-aware
switches to misbehave, e.g., by disrupting multicast packets to us.

Since the multicast snooping feature still functions without sending
our own queries, this patch will change the default to not send
queries.

For those that need queries in order to speed up convergence on start-up,
a toggle is provided to restore the previous behaviour.
"

So maybe this igmp queries conflict is the source of the problem I had.


But it seem difficult to disable it from linux bridge, if igmp snooping group could be created on physical switchs.

@Dietmar

Maybe the upgrade should be done like this:

- disable igmp snooping on physical switches (if user use it)

- upgrade kernel of each node and reboot them. (they should communicate with old and newer kernel)

- re-enable igmp snooping on physical switch when all nodes are upgraded

----- Mail original ----- 

De: "Michael Rasmussen" <mir at datanom.net> 
À: pve-devel at pve.proxmox.com 
Envoyé: Vendredi 8 Mars 2013 19:45:33 
Objet: Re: [pve-devel] corosync, multicast problem because of vmbr multicast_snooping enabled 

On Fri, 08 Mar 2013 19:34:56 +0100 (CET) 
Alexandre DERUMIER <aderumier at odiso.com> wrote: 

> 
> So, how can do you multicast with linux without bridge ? 
> Like simply use multicast from ip on eth0 ? 
> (Because it doesn't send igmp queries in this case) 
> 
Again I only presume. Given only a psychical nic I guess multicast is 
by default carried unless explicitly stated otherwise. 

-- 
Hilsen/Regards 
Michael Rasmussen 

Get my public GnuPG keys: 
michael <at> rasmussen <dot> cc 
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xD3C9A00E 
mir <at> datanom <dot> net 
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE501F51C 
mir <at> miras <dot> org 
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE3E80917 
-------------------------------------------------------------- 
Test input for validity and plausibility. 
- The Elements of Programming Style (Kernighan & Plaugher) 

_______________________________________________ 
pve-devel mailing list 
pve-devel at pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel