multicast packetloss

[Chris1]

I believe my issue is related to xos0053644, but info on that specific issue seems to be limited to https://extremeportal.force.com/ExtrArticleDetail?an=000075074 no mention in release notes etc. Unless i'm missing a way to search bulk release notes/version history for bug numbers.

Also not 100% sure this is the issue, we do seem to have issues periodically with OSPF at some sites, but i'm unsure which of my switches is the root cause.

Our layout is
Remote Site Switch (ospf x450/x460-g2 etc) -> FIBER -> MLAG-Aggregation (l2-transit x670) -> FIBER -> MLAG-Core switches (ospf x670-g2)

The one page I found says "temporary/short outages" on ospf but honestly we've seen outages of many hours or days on ospf for some sites, and it doesn't happen to all the sites.

Should i just disable the to-cpu on all ports of our aggregation switches? Is their a draw back to doing that if those switches are only qinq and vlans no l3 beyond inband management ip? Can i do the commands on a live network or will it affect traffic flow on the transit switch?

I'm having issues understanding the problem, theirs 4 solutions listed but no explanation really of figuring out which switches are the issue actually, or what the draw back is to each of the solutions

None of the options seem to be for running on the actual OSPF switch (the ones with the ip interfaces) so is the issue only on switches that are layer 2 transit switches?

[Balaji]

The hellos from the router which is in Init state is not reaching it's neighbor. but it sees the hello from it's neighbor. you need to track down the path from the init router to it's neighbor and run "show igmp snooping vlan vlan-name" and look for 224.0.0.5 entry in the senders list and find where in the path you miss the entry from one of the neighbors.

[Chris1]

An example right now we have one of our sites that is showing 1 neighbor in FULL the other neighbor doesn't show up at all.

The 2 Core switches part of same area and running vrrp+ospf on the vlan.

One of them shows FULL/FULL for both neighbors, the other core switch shows INIT/FULL

So the core switches seem to be talking to each other fine, but randomly one of the switches refuses to bring up the ospf and is stuck in init towards the remote site.

[Balaji]

Chris,

you could configure to-cpu off on a switch which is completely l2 and it is not going to participate in any layer 3 protocols.

we need to have a better understanding of the behavior to suggest a solution.

Could you explain in detail what is the issue you are observing in regards to OSPF?

Thanks