Due to high load after BGP instability at one of our upstreams, the routing daemon on our primary router stalled and the BGP and OSPF protocol sessions timed out. Sessions are coming up again and the network is recovering.

Update 17:20: The network is fully stable again.