10-11-2018 04:42 AM
When I do a BGP neighbor reset or a new BGP session goes to established, then the BGP routes are "inaccessible" for about 60 seconds.
This is how i looks shortly after the BGP session comes up:
BGP routing table entry for 1.2.3.4/32, version 0
BGP Bestpath: deterministic-med: med
Paths: (2 available, no best path)
Not advertised to any peer
Refresh Epoch 2
15623
10.0.0.1 (inaccessible) from 10.0.0.1 (1.1.1.1)
Origin IGP, metric 0, localpref 100, weight 5000, valid, external
rx pathid: 0, tx pathid: 0
Refresh Epoch 2
15623, (received-only)
10.0.0.1 (inaccessible) from 10.0.0.1 (1.1.1.1)
Origin IGP, metric 0, localpref 100, valid, external
rx pathid: 0, tx pathid: 0
This is how it looks after about 60 seconds:
BGP routing table entry for 1.2.3.4/32, version 3
BGP Bestpath: deterministic-med: med
Paths: (2 available, best #1, table default)
Not advertised to any peer
Refresh Epoch 2
15623
10.0.0.1 from 10.0.0.1 (1.1.1.1)
Origin IGP, metric 0, localpref 100, weight 5000, valid, external, best
rx pathid: 0, tx pathid: 0x0
Refresh Epoch 2
15623, (received-only)
10.0.0.1 from 10.0.0.1 (1.1.1.1)
Origin IGP, metric 0, localpref 100, valid, external
rx pathid: 0, tx pathid: 0
Because of that, the routes are installed 60 seconds after the BGP session goes to established. Can anyone explain me, what this 60 seconds timer is? Can I reduce the time, that the routes are installed faster in the routing table?
10-11-2018 04:51 AM - edited 10-11-2018 04:53 AM
Hi
BGP usually has a slow convergence but it also depends of the amount of prefixes received. You could try using BFD. Also you can customize the timers but it could not be recommended. The 60 seg is the keepalive timer. The inaccessible term is displayed until the route has a valid next hop.
10-11-2018 04:56 AM
Hi Julio
It's a lab setup, the router learns less then 5 prefixes. The hold time is 15 seconds. I don't think, I can tune something here.
The BGP process learns the prefixes verf fast, but I wonder, that he set them to "inaccessible" for around 60 seconds.
If this is a feature from BGP, then it's ok. But if I can tune them, it would be great.
10-11-2018 05:07 AM
Hi
Try with
neighbor <neighbor IP> timers <keepalive holddown> <minimum holddown>
10-11-2018 05:08 AM
This is actually the BGP hold timer in action. When you reset the neighbor the routes are put in hold-down to prevent flapping and instability. By default the hold timer is four times the hello timer (4 x 15s = 60s in your lab example). This is normal and should not be tuned to be too rapid in a production network, otherwise you risk instability. The usual way around this is to set up soft reconfiguration. Then you can do a soft reset rather than a full reset.
Hope this helps.
Dave
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide