all (after next-hop-self) is an option that was introduced by Cisco for partial mesh iBGP network (typically WAN/MAN).
The next-hop-self command (without all) exists since decades and was introduced to "hide" eBGP next-hops behind the border iBGP speaker using this command towards other iBGP speakers (typically route-reflectors).
So, Aruba follows standards on that part as many others do.
I may ask why you would need the "all" keyword in the context of LAN where peering to RR might be possible. May be a small deployment
without RR and trying to avoid full-mesh ?
If this is really important for you, please reach out to your local SE or innovation zone with ID so that I can push for it internally (thanks to your explanation of design/operational need).
------------------------------
Vincent Giles
------------------------------
Original Message:
Sent: Mar 02, 2022 06:43 AM
From: Morten Johannsen
Subject: ibgp next hop self
Hi Vincent.
Its a iBGP router reflecter client, and the routes is coming from iBGP neigbors, i can see in Cisco you can put a cli after the neighbor x.x.x.x next-hop-self "all"
Is it only working with eBGP router learned?
Morten
------------------------------
Morten Johannsen
Original Message:
Sent: Mar 02, 2022 06:23 AM
From: Vincent Giles
Subject: ibgp next hop self
just to make sure the behavior is understood: the command is applied between iBGP speakers, and the effect will be that eBGP learnt routes (not iBGP routes)
will get advertised with the IP address of the iBGP speaker the command has been set on.
------------------------------
Vincent Giles
Original Message:
Sent: Mar 01, 2022 07:11 AM
From: Morten Johannsen
Subject: ibgp next hop self
hi.
Im having some problems with getting our arubaos-cx 8360 switch to eat the neighbor x.x.x.x next-hop-self.
Its in a multi vrf env. when i do it from the peer side (cisco asr 1002x) it works, i can see recieved routers with the cisco as the next hop, but when i try from the aruba side it still announce the next hop ip from its neighbor.
I can't find anything in the white paper other that this.
Can it be a bug? version 10.07.010, this is in my lab setup, but i need to configure it on our production this saturday
Morten
------------------------------
Morten Johannsen
------------------------------