https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=288638
Bug ID: 288638
Summary: "kern.ipc.nmbufs limit reached" when running BGP and
OSPF over Wireguard
Product: Base System
Version: 14.3-STABLE
Hardware: amd64
OS: Any
Status: New
Severity: Affects Only Me
Priority: ---
Component: kern
Assignee: [email protected]
Reporter: [email protected]
I'm running FreeBSD 14.3 on two routers in two datacenters with BGP uplinks and
internal OSPF managed by net/bird2 and one OpenWrt router at home running OSPF
with bird2. Before using Wireguard, the two routers where connected via GRE
tunnels and everything was fine. Then I decided two replace the GRE tunnels
with Wireguard tunnels. First I replaced the tunnels to my home router. Then,
after one day, networking at both routers stopped working and the kernel log
printed "kern.ipc.nmbufs limit reached". I increased the nmbufs limit to
1000000 and first it looked like it worked. Now I replaced the tunnel between
the two FreeBSD routers and again I reached the nmbufs limit. At one router I
am now at kern.ipc.nmbclusters="4000000", the other one is now at
kern.ipc.nmbclusters="5000000" and I don't know if they run stable with this.
>From my understanding, both values are far away from normal.
In my Wireguard configuration I use the option Table = off and the peer
configuration has AllowedIPs = 0.0.0.0/0,::/0, because bird2 fills the routing
table for me.
Any ideas on what to look at? I once had a "normal" Wireguard setup, where
Wireguard did the routing and this setup just worked as intended, so I think it
has something to do with the combination of Wireguard and net/bird2.
--
You are receiving this mail because:
You are the assignee for the bug.
