NetworkManager consumes 100% CPU with large regularly-updated routing tables
Issue
- NetworkManager consumes 100% CPU with large regularly-updated routing tables
- We have two routing tables, each table has approximately 1 million IPv4 routes and 200,000 IPv6 routes. Total routes ~2.4M. We also run a custom routing protocol which regularly updates these tables at a rate of 15 changes per second (~900 changes per minute). Just having the routes present is not a problem, but when routes are updated then NetworkManager CPU usage spikes to 100% and stays there. Any
nmcli
command takes a long time to return.
Environment
- Red Hat Enterprise Linux 9.3 and earlier
NetworkManager-1.44.0
and earlier- Large routing tables with rapid routing table updates
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.