[j-nsp] eBGP neighbor link failure detection

[Andy Litzinger]

John,
 do you have any idea how to view the negotiated parameters for Graceful
Restart?  Or peak into its current state?

I tried setting traceoptions: protocols bgp group my-group traceoptions
flag graceful-restart detail
but that hasn't yielded any log messages.  perhaps negotiation only happens
at BGP session initiation.

is it fair to say that if you are directly connected to your neighbor and
that interface goes down that the expected behavior of GR is it should
abort and routes from that neighbor should immediately be removed?

-andy


On Fri, Mar 14, 2014 at 8:52 AM, Andy Litzinger <
andy.litzinger.lists at gmail.com> wrote:

> Hi John,
>  you might be spot on- graceful restart is configured for this peer and it
> does look like my side is respecting it:
> > show bgp neighbor
> <snip>
> Options: <Preference HoldTime AuthKey GracefulRestart LogUpDown PeerAS
> Refresh>
>
> I'll let you know what I find out
>
> -andy
>
>
> On Thu, Mar 13, 2014 at 7:10 PM, John Neiberger <jneiberger at gmail.com>wrote:
>
>> I've only seen something like this once and it was a graceful restart
>> issue....er...."feature" on a Cisco router, so I doubt it's the same
>> problem here. In that situation, the router waited for a graceful restart
>> timer to expire before it would clear out the routes, and it was doing this
>> despite the fact that it had never received a signal that the other side
>> was doing a graceful restart.
>>
>> It seems unlikely that you'd be seeing the same type of problem, but the
>> symptoms sound very similar.
>>
>> John
>>
>>
>> On Thu, Mar 13, 2014 at 3:38 PM, Andy Litzinger <
>> andy.litzinger.lists at gmail.com> wrote:
>>
>>> One of my providers (and eBGP neighbor) recently had a hardware failure
>>> which caused the port that connects our two routers to go down.  My
>>> router
>>> did detect the link failure and BGP pretty much immediately transitioned
>>> to
>>> an Idle state.  my side is a Juniper MX80  running 11.4, their side I
>>> believe is a Cisco 6509.
>>>
>>> what surprised me is that it looks like routes toward that provider were
>>> not immediately removed from my routing table.  Instead i see evidence of
>>> blackholing for almost 3 minutes.  Is this normal behavior?  is the
>>> behavior configurable? I thought it might be related to the BGP holdtime,
>>> but that appears to be set for 30 seconds.
>>>
>>>   I see that cisco has a feature called 'fast-external-fallover' that
>>> bypasses the hold-down timer.  Is there an equivalent in JunOS?  what is
>>> the Juniper best practice to handle link failure between eBGP neighbors?
>>>
>>> thanks!
>>>  -andy
>>> _______________________________________________
>>> juniper-nsp mailing list juniper-nsp at puck.nether.net
>>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>>>
>>
>>
>