Are you sure you need to slow down convergence?
I also 5 Full view eBGP sessions and with several client which receives full view from me and about 8 non full view peers.
In case of many bgp sessions flapping CPU also reach 100% load but no more than 5-7 minutes, but this is RE-1800.
This doesn't leads to problems with service.
Maybe you have some another problem.
Nevertheless I think you should look at this
https://www.juniper.net/documentation/en_US/junos/topics/topic-map/bgp-flap-damping.htmlMaybe this helps.
Original Message:
Sent: 11-19-2021 06:58
From: Unknown User
Subject: BGP Flaping session - how to prevent of high load RE CPU
Hello!
We have one chassis MX960 with installed two old RE-2000-S and several BGP sessions on it (upstream and downstream)
For our clients we advertise full view routing table with direct eBGP sessions
Now we are faced with the problem of CPU overloading at the time of transferring routes prefixes to our clients after BGP session start flapping
We fully understand that we need to swap that RE but today we are need to prevent impact of flapping BGP sessions to our CPU
The dump mechanism protects only incoming routes, but what about the routes outgoing from us?
How to set a penalty to export routes and prevent high CPU of BGP flapping session?
Thanks in advance for any answers!