Hey Ondrej, Good point... Perhaps we can specify an instance ID (which doesn't appear to be used in the BGP proto anywhere) to differentiate dynamic instances from static instances? (See patch)
Thanks, Liam On Mon, Sep 16, 2019 at 9:46 AM Ondrej Zajicek <[email protected]> wrote: > On Sun, Sep 15, 2019 at 07:58:07PM -0400, Liam Nattrass wrote: > > Hey all, > > > > I was doing some work with dynamic neighbors in BGP, and found that I am > > not able to use multiple protocol definitions with neighbor ranges. > > Connections to the second and subsequent definitions are rejected. > > > > The BGP protocol for the subsequent instances depend on a lock, but due > to > > the remote_addresses being the same (null) the subsequent protocols hang, > > waiting for the first protocol. > > > > Attached is a small patch which checks if the protocol is dynamic, and > uses > > the remote_range prefix instead of a zero remote_address for the lock. > > Hi > > Thanks for the patch. You are right about the bug. The patch fixes it, > but also breaks the case where a neighbor uses the first address from the > range (the address that is the same as the range prefix). I will check > what can be done with that. > > -- > Elen sila lumenn' omentielvo > > Ondrej 'Santiago' Zajicek (email: [email protected]) > OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) > "To err is human -- to blame it on a computer is even more so." >
0001-Support-multiple-dynamic-neighbors.patch
Description: Binary data
