From: Volkov, Dmitry (Toronto - BCE) (dmitry_volkov@ca.ml.com)
Date: Tue Sep 03 2002 - 23:38:06 GMT-3
> -----Original Message-----
> From: sean@ttank.com [mailto:sean@ttank.com]
> Sent: Tuesday, September 03, 2002 4:45 PM
> To: Volkov, Dmitry (Toronto - BCE)
> Cc: ccielab@groupstudy.com; dmitry_volkov@ca.ml.com; emonk@att.net
> Subject: RE: local policy routing
>
>
>
> Sequence is the key here.
> As you type in ping cmd to R12's loopback, nothing comes as
> the closest
>
> source interface.
>
> So, using the logic we talked about like below, IOS happens
> to source the icmp packets from R9's ehter0.
Sean,
Why does this logic exist ?? - This is the question.
Why it never sources from far end serial interfaces for example,
and why ethernet and after that loopback ?
Maybe there are any other sequences and interfaces involved...
Maybe answer is just "why not?"
Dmitry
> Only then happens the local policy routing,
> which will
>
> throw the packets
>
> out of R9's serial0 onto R12's serial0. In other words, when
> the default
>
> next-hop policy kicks
>
> in, the icmp packets are already well formed and sourced out of R9's
>
> ehter0.
>
> When the packets get to R12, we all know how and why they end up their
>
> journey at the
>
> packet bucket.
>
> HTH
>
> - Sean Liu
>
> CCIE, CCNP, CCDP, CCSE,
> AIX-CATE, CNE, MCSE+I
>
> Think Tank Systems, LLC
>
>
>
>
>
> |--------+------------------------->
> | | "Volkov, Dmitry|
> | | (Toronto - |
> | | BCE)" |
> | | <dmitry_volkov@|
> | | ca.ml.com> |
> | | |
> | | 09/03/2002 |
> | | 04:34 AM |
> | | |
> |--------+------------------------->
>
> >-------------------------------------------------------------
> -----------------------------------------------------------|
> |
> |
> | To: "'sean@ttank.com'" <sean@ttank.com>
> |
> | cc: ccielab@groupstudy.com, emonk@att.net
> |
> | Subject: RE: local policy routing
> |
>
> >-------------------------------------------------------------
> -----------------------------------------------------------|
>
>
>
> Sean,
>
> >
> > Dmitry,
> >
> > What you thought is still true, that is a router will source
> > packets from IP of interface closest to destination.
> > What happens here is, when you ping loopbacks on R12 from R9,
> > neither eth0 nor ser0 is closer to the
> > destinations (7.7.7.7 or 160.160.160.160),
>
> Right, but we have "default" next-hop configured in route-map and
> Serial on R9 is closest to next-hop.. Why it wouldn't source
> from there ?
> ;)
> You can say - this logic is by design. Ok, I'm just wondering
> what is logic
> (sourcing packets) exactly ?
>
> > by default, IOS sources packets from LAN interfaces.
>
> I would add the following: It sources from LAN interfaces
> with lowest IP.
> If there are no LAN interfaces , it sources from Loopback
> with lowest IP.
> If there are no loopbacks or ethernet, but several Serial interfaces
> it sources from serial closest to destination.
> Do You know why ? and where this behavior (dependance on int type/IP)
> described ??
>
> > Then local policy routing kicks in and sends it on to the default
> > next hop at 50.100.1.12 on R12's ser0. What really fails
> the pings is
> > because R12 does not have a return route to R9's eth0 network.
>
> Yes, I understand this.
>
> Thanks You,
>
> Dmitry
>
>
> > - Sean Liu
> >
> > CCIE, CCNP, CCDP, CCSE,
> > AIX-CATE, CNE, MCSE+I
> >
> > Think Tank Systems, LLC
> >
> >
> >
> >
> >
> > |--------+------------------------->
> > | | "Volkov, Dmitry|
> > | | (Toronto - |
> > | | BCE)" |
> > | | <dmitry_volkov@|
> > | | ca.ml.com> |
> > | | Sent by: |
> > | | nobody@groupstu|
> > | | dy.com |
> > | | |
> > | | |
> > | | 09/02/2002 |
> > | | 08:45 PM |
> > | | Please respond |
> > | | to "Volkov, |
> > | | Dmitry (Toronto|
> > | | - BCE)" |
> > | | |
> > |--------+------------------------->
> >
> > >-------------------------------------------------------------
> > -----------------------------------------------------------|
> > |
> > |
> > | To: "'Edward Monk'" <emonk@att.net>
> > |
> > | cc: ccielab@groupstudy.com
> > |
> > | Subject: RE: local policy routing
> > |
> >
> > >-------------------------------------------------------------
> > -----------------------------------------------------------|
> >
> >
> >
> > Yes Edward,
> >
> > On both sides route-map will work.
> > That I know..
> > I always thought that router sources packets from IP of
> > interface closest
> > to
> > destination.
> > Looks like it depends on type of interface when we have local policy
> > enabled
> >
> > Thanks,
> >
> > Dmitry
> >
> > > -----Original Message-----
> > > From: Edward Monk [mailto:emonk@att.net]
> > > Sent: Monday, September 02, 2002 11:37 PM
> > > To: 'Volkov, Dmitry (Toronto - BCE)'; ccielab@groupstudy.com
> > > Subject: RE: local policy routing
> > >
> > >
> > > Dmitry,
> > >
> > > You need a route back. If you apply the route-map to both
> > > sides it will
> > > work. The route-map matches all packets so no changes needed.
> > >
> > >
> > > -----Original Message-----
> > > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On
> > > Behalf Of
> > > Volkov, Dmitry (Toronto - BCE)
> > > Sent: Monday, September 02, 2002 7:28 PM
> > > To: 'ccielab@groupstudy.com'
> > > Subject: local policy routing
> > >
> > > Hello group,
> > >
> > > Could You please explain me what is wrong here R9(s0)----(s0)R12:
> > > (I can't ping loopbacks on R12 from R9)
> > >
> > > R9#
> > > interface Ethernet0
> > > ip address 170.240.8.1 255.255.255.0
> > > !
> > > interface Serial0
> > > ip address 50.100.1.9 255.255.255.0
> > > no fair-queue
> > > !
> > > ip local policy route-map test
> > > ip classless
> > > no ip http server
> > > !
> > > route-map test permit 10
> > > set ip default next-hop 50.100.1.12
> > >
> > > sh ip ro
> > > 170.240.0.0/24 is subnetted, 1 subnets
> > > C 170.240.8.0 is directly connected, Ethernet0
> > > 50.0.0.0/24 is subnetted, 1 subnets
> > > C 50.100.1.0 is directly connected, Serial0
> > > r9#
> > >
> > >
> > > R12#
> > > !
> > > interface Loopback0
> > > ip address 7.7.7.7 255.255.255.0
> > > !
> > > interface Loopback1
> > > ip address 160.160.160.160 255.255.255.0
> > > !
> > > interface Serial0
> > > ip address 50.100.1.12 255.255.255.0
> > > !
> > > ip classless
> > > no ip http server
> > >
> > > Thanks,
> > >
> > > Dmitry
> > > _________________________________________________________________
> > > Commercial lab list:
http://www.groupstudy.com/list/commercial.html
> > Please discuss commercial lab solutions on this list.
> _________________________________________________________________
> Commercial lab list: http://www.groupstudy.com/list/commercial.html
> Please discuss commercial lab solutions on this list.
This archive was generated by hypermail 2.1.4 : Mon Oct 07 2002 - 07:43:43 GMT-3