[j-nsp] MX960 NSR Problem

[Harry Reynolds]

>From the docs. Based on distributed or the timeout when not, you can have bfd timeouts during a nsr; for a gres I believe that bfd is not supported and a session reset is also expected.



When a BFD session is distributed to the Packet Forwarding Engine, BFD packets
continue to be sent during a Routing Engine switchover. If non-distributed BFD
sessions are to be kept alive during a switchover, you must ensure that the session
failure detection time is greater than the Routing Engine switchover time. The
following BFD sessions are not distributed to the Packet Forwarding Engine: multi-hop
sessions, tunnel-encapsulated sessions, and sessions over aggregated Ethernet and
Integrated Routing and Bridging (IRB) interfaces.

NOTE: For BFD sessions to remain up during a Routing Engine switchover event
when nonstop active routing is configured, the value for the minimum-interval
configuration statement (a BFD liveness detection parameter) must be at least 2500
ms for Routing Engine-based sessions and at least 10 ms for distributed BFD sessions.



-----Original Message-----
From: juniper-nsp-bounces at puck.nether.net [mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of Walaa Abdel razzak
Sent: Monday, July 13, 2009 10:54 AM
To: Richard A Steenbergen
Cc: juniper-nsp at puck.nether.net
Subject: Re: [j-nsp] MX960 NSR Problem

OK, That's why I see BGP session reset on the new Backup RE but not on the new master because of resync process and we can say this behaviour is normal. But why the BFD, ISIS sessions was reset on the new master and affected neighbours state?

Best Regards,
Walaa Abdel Razzak

-----Original Message-----
From: Richard A Steenbergen [mailto:ras at e-gerbil.net]
Sent: Monday, July 13, 2009 12:55 AM
To: Walaa Abdel razzak
Cc: juniper-nsp at puck.nether.net
Subject: Re: [j-nsp] MX960 NSR Problem

On Sun, Jul 12, 2009 at 08:36:12PM +0300, Walaa Abdel razzak wrote:
> Experts
> 
> I have MX960 router with JUNOS 9.4R2 having GRES and NSR enabled. When
I
> switch from the master to backup RE, Traffic is not interrupted but
the
> protocols appears to reset on the old master but the strange thing is 
> that it appears not reset on the new master, do you have reason for 
> this? The BGP on the neighbor routers didn't flap, but ISIS, BFD
session
> flapped on the neighbors.

That is how NSR is supposed to work. When running NSR, the backup RE is pre-synchronized to the state of the active RE, so when you do a switchover the BGP session appears to be uninterrupted. At some point after the switchover is completed, the new backup RE should sync up in the same way, but it won't be instant.

BTW if I were you I'm not sure that I would be running NSR in production on 9.4R2. I've seen a lot of bugs related to it (and even related to GRES, ever since around 9.1+ it has been very sketchy for me) which can cause disruptions or even blackholing within packet forwarding when "something" happens on the backup RE. LDP sessions resetting whenever the backup RE was rebooted was extremely annoying. :)

--
Richard A Steenbergen <ras at e-gerbil.net> http://www.e-gerbil.net/ras GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1
2CBC)
 

__________ Information from ESET Smart Security, version of virus
signature database 4223 (20090708) __________

The message was checked by ESET Smart Security.

http://www.eset.com
 
 

__________ Information from ESET Smart Security, version of virus
signature database 4223 (20090708) __________

The message was checked by ESET Smart Security.

http://www.eset.com
 
_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp