Created attachment 470430 [details] See "Additional info" in the initial comment. Description of problem: With the late kernels, bringing up an Ethernet bridge enables multicast routing and IGMP snooping for that bridge (and its associated interfaces). This leads in a lot of IGMP snooping packets emitted on the pertaining interfaces. Unfortunately, the pseudo-files controlling this feature are not located in procfs, but in sysfs (/sys/classes/net/${device}/bridge/multicast_*), for which there is neither a tool nor a structure similar to sysctl.conf nor some device-dependent configuration parameters. Version-Release number of selected component (if applicable): initscripts-9.20.1-1.fc14.x86_64 How reproducible: Always Steps to Reproduce: 1. Create and activate a bridging interface. 2. Watch the connected network for packets with dest.ip = 224.0.0.1 and protocol=2 Actual results: Some multicast packets are seen. Expected results: Some way to disable the emission of these packets. Additional info: The attached patch introduces MULTICAST_* bridge configuration parameters matching the multicast pseudo-file names. These parameters can be defined in the ifcfg-br? files for fine control. Although all multicast_* files can be set this way, it is particularly useful for MULTICAST_ROUTER and MULTICAST_SNOOPING. The patch translates non-numeric values as: (case insensitive) "yes"-->1, everything else as 0. After applying this patch, setting up MULTICAST_SNOOPING="no" in ifcfg-br0 of my test system and a restart, IGMP snooping multicast packets are no longer seen on my network.
Ick. I generally dislike the idea of unbounded sets of configuration variables like this being defined (and then having to worry about migrating configs if they change, etc.) Much like with newer random opts for ethtool, and so on, I believe these sorts of things are likely better configured with udev rules. Failing that, a more generic approach such as what's used by BONDING_OPTS might be better.
udev is hard and hazardous to configure for a "regular" sysadmin (in fact, I do not know how to do it this way!). In addition, I like the idea of having all parameters for a given net device in a single config file... If you dont like my patch, please consider this report as a bug, or at least as a missing feature that must be fixed/provided. The reason is: if you have several computers with bridged interfaces on the same lan, these packets generate important network and firewall log pollution. If you're OK to apply a patch "à la" BONDING_OPTS and want me to rewrite it, please ask. Thanks.
If I might chime in, as someone that has the pleasure of working with machines with large number of NIC's and bridges I would prefer having this in ifup and not in udev, as it's far easier for me to instruct the system administrators to disable it within the ifcfg-* files they already know and use.
Created attachment 520676 [details] potential patch What do you think of the above?
Looks great. - Gilboa
I did not test it, but if it works for you, it certainly fulfils my needs. Thanks
Added as http://git.fedorahosted.org/git/?p=initscripts.git;a=commitdiff;h=84d0dadb52b1dfb4ecb01d1416c59da6a8af166c Will be in F-16 and later.
Thanks a lot.
initscripts-9.32-1.fc16 has been submitted as an update for Fedora 16. https://admin.fedoraproject.org/updates/initscripts-9.32-1.fc16
Package initscripts-9.32-1.fc16: * should fix your issue, * was pushed to the Fedora 16 testing repository, * should be available at your local mirror within two days. Update it with: # su -c 'yum update --enablerepo=updates-testing initscripts-9.32-1.fc16' as soon as you are able to. Please go to the following url: https://admin.fedoraproject.org/updates/initscripts-9.32-1.fc16 then log in and leave karma (feedback).
initscripts-9.32-1.fc16 has been pushed to the Fedora 16 stable repository. If problems still persist, please make note of it in this bug report.