[c-nsp] SSO with ISIS NSF (IETF vs Cisco)

Tue Jan 15 12:18:12 EST 2008

Atif,

Thanks for the config.

The problem isnt in ISIS, its LDP.  MPLS SSO and LDP graceful-restart
are not
supported in 12.0S.  The command you have in your config "mpls ldp
graceful-restart"
only enables LDP helper mode, allowing the router to act as a peer to a
switching-over router.

Given this, RPR+ like behavior is expected.  I also suspect the NSR
version of ISIS
allows LDP to come up faster, which is why you see better times in that
scenario.

MPLS SSO is only supported on the 12000 in IOS XR.

Hope this helps,
Ken

________________________________

From: Ken Weissner (kweissne) 
Sent: Tuesday, January 15, 2008 11:35 AM
To: 'Atif Sid'
Cc: cisco-nsp at puck.nether.net
Subject: RE: [c-nsp] SSO with ISIS NSF (IETF vs Cisco)

Can you provide the complete config and a sh hardware for P2 (offlist if
necessary)
and I'll see what I can tell from them.

There should be no difference in switchover behavior between different
RP slots.

Ken

________________________________

From: Atif Sid [mailto:guru6111 at gmail.com] 
Sent: Tuesday, January 15, 2008 11:26 AM
To: Ken Weissner (kweissne)
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] SSO with ISIS NSF (IETF vs Cisco)

another interesting this i just noticed is that

Failover from RP8 to RP9 takes 30 Sec
while  from RP9 to RP8 is 8 secs

could there be config sync issue?

FAILOVER FROM RP8 TO RP9: (it reloads)

P2#redundancy force-switchover 
Proceed with switchover? [confirm]
System Bootstrap, Version 12.0(20040128:214555) [assafb-PRP1P_20040101
1.8dev(2.83)] DEVELOPMENT SOFTWARE
Copyright (c) 1994-2004 by cisco Systems, Inc. 

DRAM DIMM Slot 1: 2048M found, Slot 2: Empty 
MPC7457 platform with 2097152 Kbytes of main memory

Self decompressing the image :
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
BBBBBBBBBBBBBBBBBBBBBBB [OK] 

while RP9 to RP8:

P2#redundancy force-switchover 
Proceed with switchover? [confirm]
Switching roles - Standby to Active

SEC  8:*Jan 15 11:30:47.913 EST: %MBUS-6-SWITCHOVER: Switchover
initiated by active in slot 9, management request
SEC  8:*Jan 15 11:30:47.953 EST: %RP-5-NEWPRIMARY: Switchover to new RP
*Jan 15 11:30:47.973 EST: %LINK-3-UPDOWN: Interface Ethernet0, changed
state to up 
*Jan 15 11:30:47.977 EST: %LINK-5-CHANGED: Interface Ethernet1, changed
state to administratively down
*Jan 15 11:30:47.981 EST: %LINEPROTO-5-UPDOWN: Line protocol on
Interface , changed state to up
*Jan 15 11:30: 47.985 EST: %TAGCON-3-LCLTAG_ALLOC: Cannot allocate local
tag
*Jan 15 11:30:48.974 EST: %LINEPROTO-5-UPDOWN: Line protocol on
Interface Ethernet0, changed state to up
*Jan 15 11:30:48.978 EST: %LINEPROTO-5-UPDOWN: Line protocol on
Interface Ethernet1, changed state to down 
*Jan 15 11:30:48.982 EST: %LINEPROTO-5-UPDOWN: Line protocol on
Interface Ethernet2, changed state to down
*Jan 15 11:30:49.610 EST: %MBUS-6-RP_STATUS: RP in Slot 8 Mode = MBUS
Active

and i see sessions are UP within 8 seconds. 

IETF were both 30 Seconds.

On 1/15/08, Atif Sid <guru6111 at gmail.com> wrote: 

	Ok. and this was what I exptected but testing shows that when I
use IETF mode and failover it takes ~ 30 sec  + to  repair while in
Cisco mode it can converge in 8 seconds.

	Here is a test:

	3 routers: cisco 12410/PRP, IOS: 12.0(32)SY3

	P1 --> P2 --> P3

	* Pinging from p1 to p3 with timout set to 1 sec , all three
routers are in NSF Cisco. 
	* force a RP failover on P2, 

	P1#ping
	Protocol [ip]: 
	Target IP address: p3
	Repeat count [5]: 10000
	Datagram size [100]: 
	Timeout in seconds [2]: 1
	Extended commands [n]: 
	Sweep range of sizes [n]: 
	Type escape sequence to abort. 
	Sending 10000, 100-byte ICMP Echos to x.x.x.x, timeout is 1
seconds:

	Success rate is 99 percent (9992/10000), round-trip min/avg/max
= 1/1/4 ms
	P1#

	On 1/15/08, Ken Weissner (kweissne) <kweissne at cisco.com > wrote:

		Hello,

		I've just joined this list, so didnt have a copy of the
original email
		to reply to, which was forwarded by a colleague. 

		>My observation is that Cisco mode is faster then IETF,
in fact IETF
		has as delay as being in RPR+ mode?
		>Is IETF a GR? vs Cisco a SSO?

		There should be no difference in length of outage if
both these modes 
		are functioning properly.
		IETF is GR mode, ie RFC3847.   "Cisco" mode is a
"Non-Stop Routing" mode
		that doesnt require
		interaction with the peer to recover from the processor
failure due to 
		the transfer of all state information to the standby.

		A "RPR+" like delay could be that the peer to the
switching over router
		doesnt have helper capability.

		Thanks,
		Ken Weissner

		_______________________________________________ 
		cisco-nsp mailing list  cisco-nsp at puck.nether.net
		https://puck.nether.net/mailman/listinfo/cisco-nsp
		archive at http://puck.nether.net/pipermail/cisco-nsp/ 