Which BGP timer has a default value of 30 minutes?

[d.wingert]

Hello Everyone,

Observation: After applying the cmd 'bgp bestpath med missing-as-worst' on a router which already has "bgp bestpath as-path ignore" configured, a 'show ip bgp' shows prefixes which didn't have a MED assigned previously with a new value of "4294967295" - which is what I expected to see.

What I didn't expect was that the router continued to use the paths with the MED value of "4294967295" for a full 30 minutes before changing to paths which had lower MED values.

Q: Any idea which timer within BGP is responsible for this behavior?

This is in a lab environment which physical hardware, running c2800nm-adventerprisek9-mz.151-4.M7

• Only routing protocol used was eBGP with each router peering only with its directly connected neighbors:

R1 (AS100) - R2 (AS200)

 |                            |

R4 (AS400) - R3 (AS300)

• Each router only advertises a single /8 prefix into BGP (i.e. R1= 1.0.0.0/8, R2= 2.0.0.0/8, R3= 3.0.0.0/8, R4= 4.0.0.0/8)
• R1 has 'bgp bestpath med missing-as-worst' & 'bgp bestpath as-path ignore' configured.
• R4 has an out-going route-map pointing at R1 which sets the MED to 100 for all prefixes
• R2 doesn't modify any attributes so the prefixes 3.0.0.0/8 and 4.0.0.0/8 will show up on R1 with the MED value of '4294967295' once 'bgp bestpath med missing-as-worst' is applied to R1
• No changes where where performed during the 30 minutes, R1 just dynamically changed to the paths with the lower MED values after 30 minutes.

Thanks for your help!

Dirk.

[milan.kulik]

Hi,

did  'show ip bgp' still show the paths with MED  '4294967295' as the best ones until those 30 minutes passed?

If you issue "clear ip bgp * soft" on R1 after 'bgp bestpath med missing-as-worst' applied, does it still take 30 minutes to move to the paths with lower MED?

BR,

Milan

[d.wingert]

Hi Milan,

Yes, 'show ip bgp' continued to show the paths with the MED of '4294967295' as the best until the 30 minutes passed.

Soft clears (in and out) on R1 as well as it's peers (R2 and R4) didn't have any effect.

Only a hard reset via "clear ip bgp *" causes R1 to choose paths with a lower MED after the peerings re-established.

Thanks,

Dirk

[Joseph W. Doherty]

Soft clears (in and out) on R1 as well as it's peers (R2 and R4) didn't have any effect.

Both those routers support soft resets?

[d.wingert]

Hello Joseph,

Yes all four routers in the topology support it and I tried it on R1 and its neighbors: R2 and R4.

IOS on the devices: c2800nm-adventerprisek9-mz.151-4.M7

Thanks,

Dirk

[paul driver]

Hello
Please review this excellent documentation on BGP convergence and it scanner process - Here

res
Paul

[paul driver]

Hello
Try enabling bgp always-compare-med also and as suggested perform a soft reset

res

Paul

[d.wingert]

Hi Paul,

I will revisit the scenario and will report back.(soft resets on R1 and its peers without the "bgp always-compare-med" had no effect)

My main interest though is in which timer is at play here, possible that this can only be answered by a Cisco employee familiar with the inner workings of the IOS as I haven't been able to find any BGP related timer with a default value of 30 minutes/1800 seconds so far.

It might even be IOS version dependent. (will explore that as well)

Thanks,

Dirk