Re: eigrp SIA queries

SIA problem only affects certain prefixes - the ones that router can't
receive all replies for. If you want to reset SIA prematurely, use the
command "show ip eigrp topology active" to find the active prefixes
and neighbors you still awaiting responses from (marked by the "r"
flag) and clear these neighbors manually, using the command "clear ip
eigrp neighbors". However, this will not solve the root cause, just
clears the symptoms.

You goal, when troubleshooting SIA problems, is finding the
misbehaving router or link in the network. Normally, you may want to
use the above mentioned command "show ip eigrp topology active" to
trace the router that is constantly failing to send a response.
Commonly, this is related to link problems, such as heavy congestions,
noisy or low bandwidth links, which is often combined with a high
value for the EIGRP "hold" timer (this timer affect EIGRP RTP
retransmissions). You primary indicator of SIA problem, which could
prompt you to start researching the issue, is EIGRP syslog message
"%DUAL-3-SIA" registering SIA condition.

Since IOS 12.1(something) EIGRP implements a very helpful SIA rewrite
feature. In short, using additional signaling, this feature allows
pushing neighbor resets close to the misbehaving router, as opposed to
clearing neighbor relations at the query source. Therefore, if you
have "eigrp log neighbor-changes" enabled in your network (which is
the default) you may quickly track the router that is resetting its
neighbor due to "retry limit exceeded" or "DUAL SIA" conditions. This
specific neighbor/link is most likely the source of the problem.

The general solution to the SIA problem is proper network design:
effective prefix summarization along with of EIGRP stubs.
Additionally, EIGRP design may require proper tuning of interface
bandwidth values combined with careful selection of EIGRP hold timers
on slow links. You should remember that EIGRP is a fragile protocol,
where convergence failure probability grows linearly with the query
scope range. In large networks that dont utilize proper summarization
a query may leak to slow remote spoke and cause SIA conditions many
hops aways from the origin of the query.

-- 
Petr Lapukhov, petr_at_INE.com
CCIE #16379 (R&S/Security/SP/Voice)
Internetwork Expert, Inc.
http://www.INE.com
Toll Free: 877-224-8987
Outside US: 775-826-4344
2010/5/27 ehtesham ali <conect2ehtesham_at_gmail.com>:
> HI gurus,
> i needed clarification on the following questions
>
> 1)  router R1 running eigrp is struck in SIA . after three min R1 lost
> neighbor relationship with R2 . how do i make R1 & R2 come out of SIA ..i
> can think of rebooting the routers . but is there any command similar to
> CLEAR IP OSPF PROCESS in Eigrp ?
>
> whats the best way to make routers come out of SIA without rebooting them ?
> can i reset and restart the eigrp process again ?
>
> or is it that the routers became neighbors automatically when SIA stops say
> after 5 min ?
>
> 2) if successor is down then traffic is routed via FS . After some time
> successor comes online again .will the traffic now routed thorough successor
> ? or it will continue to send through FS .  (i mean is preemption  is
> available ? )
>
> thanks
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
Blogs and organic groups at http://www.ccie.net