[c-nsp] BGP Hold Time Expired, but why?

Ozgur Guler gulerozgur at yahoo.co.uk
Tue Jul 22 08:06:09 EDT 2008 

When you do "show ip bgp nei detail"  while the sessions are flapping, 
Do you see anything wrong in the TCP parameters? 
I remember a bug in 12.0S where TCP window size becomes 0 for BGP causing it to flap. Or if it is an MTU problem you might see that the BGP Keepalives are being throttled. 

--- On Tue, 22/7/08, Christian Koch <christian at broknrobot.com> wrote:
From: Christian Koch <christian at broknrobot.com>
Subject: Re: [c-nsp] BGP Hold Time Expired, but why?
To: "Ozgur Guler" <gulerozgur at yahoo.co.uk>
Cc: "Oliver Boehmer (oboehmer)" <oboehmer at cisco.com>, "cisco-nsp" <cisco-nsp at puck.nether.net>
Date: Tuesday, 22 July, 2008, 12:54 PM

they are all 7609-S

On Tue, Jul 22, 2008 at 5:25 AM, Ozgur Guler <gulerozgur at yahoo.co.uk> wrote:


Is this a GSR?



--- On Tue, 22/7/08, Christian Koch <christian at broknrobot.com> wrote:

From: Christian Koch <christian at broknrobot.com>
Subject: Re: [c-nsp] BGP Hold Time Expired, but why?
To: "Oliver Boehmer (oboehmer)" <oboehmer at cisco.com>

Cc: "cisco-nsp" <cisco-nsp at puck.nether.net>
Date: Tuesday, 22 July, 2008, 1:58 AM

same issue, no differences...got me





On Sun, Jul 20, 2008 at 2:53 AM, Oliver Boehmer (oboehmer) <
oboehmer at cisco.com> wrote:

>  I don't know, but I would try it..  Looks weird..

>
>     oli
>
>  ------------------------------
> *From:* Christian Koch [mailto:christian at broknrobot.com]
> *Sent:*
 Saturday, July 19, 2008 7:07 PM
>
> *To:* Oliver Boehmer (oboehmer)
> *Cc:* cisco-nsp
> *Subject:* Re: [c-nsp] BGP Hold Time Expired, but why?
>
>  config look ok as far as i can see, i actually dont have bgp router-id

> set in the bgp config... you think if i add that with the loopback ip, it
> would make a difference?
>
>
> config
>
> router bgp 65000
>  no synchronization
>  bgp log-neighbor-changes

>  bgp graceful-restart restart-time 120
>  bgp graceful-restart stalepath-time 360
>  bgp graceful-restart
>  bgp dampening
>  neighbor Backbone peer-group
>  neighbor Backbone remote-as 65000

>  neighbor Backbone update-source Loopback1
>  neighbor Backbone version 4
>  neighbor Backbone send-community
>  neighbor 10.10.10.2 peer-group Backbone

>  neighbor 10.10.10.3 peer-group Backbone
>  no
 auto-summary
>
>
>
>
>
> On Sat, Jul 19, 2008 at 12:29 PM, Oliver Boehmer (oboehmer) <
> oboehmer at cisco.com> wrote:

>
>> Hmm, "%BGP-5-ADJCHANGE: neighbor 10.10.10.3 Down BGP protocol
>> initialization" looks unexpected, not sure what's happening..
>> just a hunch, but can you double-check your config regarding loopback

>> addresses, bgp router-id and things? Possibly add some bgp debug (deb
>> bgp all events, deb bgp all, deb bgp all keep) and see if something
>> weird pops up?
>> What does the neighbor's (10.10.10.3) log say?

>>
>>    oli
>>
>> ________________________________
>>
>> From: Christian Koch [mailto:christian at broknrobot.com]

>> Sent: Saturday, July 19, 2008 3:08 PM
>> To: Oliver Boehmer (oboehmer)
>> Cc: cisco-nsp
>> Subject: Re: [c-nsp]
 BGP Hold Time Expired, but why?
>>
>>
>> hmm, i didnt check cef/mpls on the new path, i should try that.. there
>> is connectivity between the loopbacks
>>
>> the session comes back up right after the timer expires.thats what

>> puzzles me
>>
>> actually 3-4 is about how long i kept it down for..
>>
>>
>> Jul 16 14:29:22 EDT: %LINK-3-UPDOWN: Interface TenGigabitEthernet2/2,
>> changed state to down

>> Jul 16 14:29:22 EDT: %LINEPROTO-SP-5-UPDOWN: Line protocol on
Interface
>> TenGigabitEthernet2/2, changed state to down
>> Jul 16 14:29:22 EDT: %OSPF-5-ADJCHG: Process 10, Nbr 10.10.10.2 on

>> TenGigabitEthernet2/2 from FULL to DOWN, Neighbor Down: Interface down
>> or detached
>> Jul 16 14:29:22 EDT: %LDP-5-NBRCHG: LDP Neighbor 10.10.10.2:0 (11) is

>> DOWN (Interface not
 operational)
>> Jul 16 14:29:22 EDT: %LINK-SP-3-UPDOWN: Interface
TenGigabitEthernet2/2,
>> changed state to down
>> Jul 16 14:29:23 EDT: %LINK-SP-3-UPDOWN: Interface
TenGigabitEthernet2/2,

>> changed state to up
>> Jul 16 14:29:23 EDT: %LINK-3-UPDOWN: Interface TenGigabitEthernet2/2,
>> changed state to up
>> Jul 16 14:29:23 EDT: %LINEPROTO-5-UPDOWN: Line protocol on Interface

>> TenGigabitEthernet2/2, changed state to up
>> Jul 16 14:29:23 EDT: %LINEPROTO-SP-5-UPDOWN: Line protocol on
Interface
>> TenGigabitEthernet2/2, changed state to up
>> Jul 16 14:29:33 EDT: %LDP-5-NBRCHG: LDP Neighbor 10.10.10.2:0 (11) is

UP
>> Jul 16 14:30:19 EDT: %OSPF-5-ADJCHG: Process 10, Nbr 10.10.10.2 on
>> TenGigabitEthernet2/2 from LOADING to FULL, Loading Done
>> Jul 16 14:30:37 EDT: %LDP-5-NBRCHG: LDP Neighbor 10.10.10.2:0 (4)
 is
>> DOWN (Discovery Hello Hold Timer expired)
>> Jul 16 14:31:39 EDT: %LDP-5-NBRCHG: LDP Neighbor 10.10.10.2:0 (4) is
UP
>> Jul 16 14:32:38 EDT: %BGP-3-NOTIFICATION: received from neighbor

>> 10.10.10.3 4/0 (hold time expired) 0 bytes
>> Jul 16 14:32:38 EDT: %BGP-5-ADJCHANGE: neighbor 10.10.10.3 Down BGP

>> protocol initialization
>> Jul 16 14:32:45 EDT: %BGP-5-ADJCHANGE: neighbor 10.10.10.3 Up
>>
>>
>>
>> On Sat, Jul 19, 2008 at 3:24 AM, Oliver Boehmer (oboehmer)

>> <oboehmer at cisco.com> wrote:
>>
>>
>>        No clue what's happening.. I've seen issues in the past
with TCP

>> PMTUD
>>        when the path converges over a link with a different MTU (which
>> is
>>        happening in your case), but as BGP will not send packets
larger
>> than
>>      
  4k, this shouldn't be an issue here.
>>
>>        How long did you take down the link before bringing it back up?
>> I assume
>>        longer than 3 minutes? Have you checked CEF and MPLS along the

>> new path?
>>        You have IP connectivity between the loopbacks aR1 and bR2?
Does
>> the
>>        session come back up eventually, or will it stay down?
>>
>>               oli

>>
>>        Christian Koch <> wrote on Saturday, July 19, 2008 8:38
AM:
>>
>>
>>        > sorry forgot to specify
>>        >
>>        > the bgp session from aR1 to bR2 is the session in question

>>        >
>>        > ck
>>        >
>>        > On Sat, Jul 19, 2008 at 2:21 AM, Christian Koch
>>        > <christian at broknrobot.com>
 wrote:
>>        >
>>        >> Hello -
>>        >>
>>        >> I have the following topology in lab, testing
different
>> failure
>>        >> scenarios. When i disconnect the link between aR1 and

bR1,
>> what
>>        >> would appear to be normal happens - ospf and ldp
neighbor go
>> down.
>>        >>
>>        >> When i re-connect the link between aR1 and bR1, the

interface
>> comes
>>        >> back up, osfp/ldp neighbor is re-established.
>>        >>
>>        >> 3minutes later, bgp holdtime expires , and all links
are up..

>>        >>
>>        >> aR1-----------------bR1
>>        >>>                         |
>>        >>>                         |
>>       
 >>>                         |
>>        >>>                         | aR2-----------------bR2
>>        >>
>>        >>
>>        >> Some Notes

>>        >> - All Links 10GE
>>        >> - Full ibgp mesh
>>        >> - Peering is to loopbacks
>>        >> - OSPF as IGP
>>        >> - Loopbacks in OSPF

>>        >> - MPLS Enabled on Interfaces
>>        >>
>>        >>
>>        >> OSPF cost between aR1 and aR2 is 1
>>        >> OSPF cost between bR1 and bR2 is 1

>>        >> OSPF cost between aR1 and bR1 is 250
>>        >> OSPF cost betwen aR2 and bR2 is 500
>>        >>
>>        >> MTU 9216 between aR1 and aR2, aR1 and bR1, aR2 AND BR2

>>        >> MTU 9182
 between bR1 and bR2
>>        >>
>>        >>
>>        >> IOS on aR1 and aR2 is 12.2.33.SRB2 -  SUP720
>>        >> IOS on bR1 and bR2 is 12.33.SRC - RSP720

>>        >>
>>        >>
>>        >> i am stumped, any ideas would be helpful in trying to
>> understand why
>>        >> the bgp session is going down due to expired hold

time, when
>> all
>>        >> links are up..
>>        >>
>>        >> thanks!
>>        >>
>>        >> ck
>>        >>

>>        >>
>>        >>
>>        >>
>>        >>
>>        >>
>>        >
>>        >
>>        > --
>>        > ^christian$

>>
>>       
 > _______________________________________________
>>        > 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/
>>
>>
>>
>>
>>
>> --

>> ^christian$
>>
>
>
_______________________________________________
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/




       
Not happy with your email address?

  Get the one you 

really want - millions of new email addresses available now at  Yahoo!




      __________________________________________________________
Not happy with your email address?.
Get the one you really want - millions of new email addresses available now at Yahoo! http://uk.docs.yahoo.com/ymail/new.html

More information about the cisco-nsp mailing list