[j-nsp] Output for "show bgp summary" is showing different on two REs

[Stefan Fouant]

> -----Original Message-----
> From: juniper-nsp-bounces at puck.nether.net [mailto:juniper-nsp-
> bounces at puck.nether.net] On Behalf Of saurabh sood
> Sent: Saturday, June 25, 2011 9:41 AM
> To: juniper-nsp at puck.nether.net
> Subject: [j-nsp] Output for "show bgp summary" is showing different on
> two REs
> 
> Hello,
> 
> I have observed a different scenario, i don't know wthr it is by
> expected or
> not.
> 
> Router has two REs and GRES, NSR is enabled also commit synchronize is
> there.
> But while checking the output for "show bgp summary" then on Master RE
> everything is showing ok.
> 
> On the other hand backup RE is showing all bgp states IDLE.
> 
> Is it an expected behavior?

IIRC, this is normal behavior.  You don't actually want RPD on each
respective RE attempting to form BGP adjacencies with remote peers at the
same time.

On the other hand, all the kernel state is replicated - as BGP packets come
in they are replicated to both REs so that they both have local copies of
the RIB and based on this should compute identical copies of the FIB tables,
etc. (The reason we don't simply synchronize the state of the Backup RE to
the Primary is because of Fate Sharing - if there is a problem on the
Primary you don't want the Backup to be synched and therefore have similar
problems).

I believe as soon as you perform GRES (or should the Primary fail for some
reason), the BGP finite state machine should immediately toggle from the
Idle state to the Established state (or whatever a particular peering state
happened to be on the Primary RE).

Sorry I don't have more specifics, this is covered in more detail in 'Junos
High Availability' by James Sonderegger, Orin Blomberg, Kieran Milne, and
Senad Palislamovic.  This is an excellent book if you want more details
regarding operations such as the above.

HTHs.

Stefan Fouant
JNCIE-M #513, JNCIE-ER #70, JNCI
http://www.shortestpathfirst.net
GPG Key ID: 0xB4C956EC