cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2066
Views
20
Helpful
20
Replies

Router sends packets out wrong interface

trane.m
Level 1
Level 1

Hi all,
A section of my lab:

tranem_0-1670500834637.png

I'm on the windows server, trying to get it to connect to the internet via the routers (R1) g0/0 interface. i give it the IP of 10.1.50.5/24 and i send a ping to 10.1.1.1. In my wireshark captures, i can see the packet is going all the way to R1, but also that R1 chooses to send the ICMP reply out it's g0/2 interface. In fact, i can see that traffic to all my VLANs are preferred out of the g0/2 interface. How do i make it send traffic specifically for vlans 10, 20 and 50 out it's g0/1 interface?

R1 has learned routes to all the SVI's via OSPF. I have created SVI's on both Core1 (10.1.XX.1) and Core2 (10.1.XX.2) but for some reason R1 prefers Core2 for all SVI routes. R1's route table:

tranem_1-1670501270138.png

I have tried creating a new route to 10.1.50.0/24 via g0/2 with a higher AD, but of course that doesn't work, when it already has a better route. I thought about creating another route out of it's g0/1 interface with a lower AD, but that would just result in it sending all traffic out of it's g0/1 interface. The same with OSPF cost.

I hope someone can point me in the correct direction.

1 Accepted Solution

Accepted Solutions

because you have not read my reply before I posted :

10.1.50.0 is behind CORE1 only,  it needs to learn from CORE1 only since that is the only path here.

if you looking to build that topology. below marked 

balajibandi_0-1670525467673.png

 

Core1 - G 2/3 - convert in to Layer 2 trunk

SRVACC1 - g0/0 covert in to layer 2 trunk

SRVACC1 - g0/1 - make access port vlan 50 (make sure vlan 50 created)

 

CORE1

config t

interface vlan50

no shutdown

!

CORE2

!

no interface vlan50

shutdown

!

Test :

from Server 10.1.50.5 gateway 10.1.50.1

ping 10.1.50.1 - if that works, then 10.1.1.1 should work as expected

still issue post new show from all device to review.

BB

***** Rate All Helpful Responses *****

How to Ask The Cisco Community for Help

View solution in original post

20 Replies 20

will run small lab and check one idea I have. 
I will update you soon 

trane.m
Level 1
Level 1

If i change the IP of SRVACC1's G0/1 int to 10.1.4.1 and the IP of the windows server to 10.0.4.2, the windows server can ping the router. So i guess it must have something to do with the configuration of the SVI's and the routes, but im not sure.

I am out but 
I think the cores run HSRP ?
if yes then 
ip sla toward the VIP of Core HSRP 
then config two static route one with high AD 
other with default AD and track with IP SLA of VIP of HSRP 
this make traffic send via active Core and router use interface to active core for return traffic

The cores run GLBP. Here is a snippet of how the SVI's are configured (this is from Core 2):

tranem_0-1670503783889.png

 

 

two solution here 
1-  config NAT, this way the Router will reply to IP of link connect to Core that send the traffic 
2- change GLBP to be HSRP since GLBP have active/active but HSRP have active/standby 

Can you explain to me what the problem is?

the router can have two equal static route 
this give router two path one via g0/1 and other g0/2. 

Core run GLBP which is active active 

now assume Core1 forward traffic to Router 
router will select path either through g0/1 (it OK) or through g0/2 that is asymetric routing 

how we can solve this 
1- NAT 
instead of router see your LAN subnet it will see IP of Link between Core and router 
so the router will always select the nearest interface to destination (for return back traffic)
2- using HSRP, because the HSRP work as active/standby 
but wait how we can force router to answer through right interface ?
by using IP SLA track as I mention above. 

all of this is to eliminate the asymetic routing.

I'm not sure if i know enough about natting to implement it.

i would also be sorry to ditch GLBP, because i think it is an interesting protocol with effective feature, like load balancing.

I've never heard of IP SLA track (i'm a student) so i will have to look that up. Thank you for your time

balaji.bandi
Hall of Fame
Hall of Fame

 

balajibandi_0-1670506218921.png

can you post

show ip route

from CORE 1 and CORE2 (along with config) how is your OSPF point to point ?

10.1.50.0 is behind CORE1 only,  it needs to learn from CORE1 only since that is the only path here.

simple lab :

balajibandi_1-1670506445820.png#

balajibandi_2-1670506464855.png

balajibandi_3-1670506491868.png

 

 

balajibandi_4-1670506516500.png

 

 

 

BB

***** Rate All Helpful Responses *****

How to Ask The Cisco Community for Help

Hello,

Config is long but attached. Same file for both core switches, so look for sections.

Core 1 route table:

tranem_0-1670508329993.png

Core 2 route table:

tranem_1-1670508360020.png

 

 

 

 

as per the orginal post you have issue with 10.1.50.X network to reach 10.1.1.1 IP address - is this issue ?

if you running OSPF, you do not need any static routing so you can remove static routes.

i am not sure how is your full topology where you looking to use other subnet - as per concern 10.1.50.X network

if you looking to set up HSRP then that is the wrong topology you created for network 10.1.50.X network - that not going to work.

 

BB

***** Rate All Helpful Responses *****

How to Ask The Cisco Community for Help

Yes, as per the original post, i have issues with a windows server sending ICMP echo requests from 10.1.50.4 to 10.1.1.1 (R1 G0/1) via switch SRVACC1, but the ICMP echo replies are sent out of R1 G0/2. I'm trying to understand why and what went wrong.

Will OSPF also find default gateway by itself?

I'm not looking to set up HSRP, i'm using GLBP. Picture attached of whole topology.

tranem_0-1670510002553.png

If i can provide any other information, please tell me. Thank you for your time.

 

Yes, as per the original post, i have issues with a windows server sending ICMP echo requests from 10.1.50.4 to 10.1.1.1 (R1 G0/1) via switch SRVACC1, but the ICMP echo replies are sent out of R1 G0/2. I'm trying to understand why and what went wrong.

 

for this to work do below steps and test it :

CORE1

config t

interface vlan50

shutdown

!

CORE2

!

interface vlan50

shutdown

!

now you test it, i will review your topolgy and give you suggestion for the GLBP for VLAN 50 to work as expected.

 

if above suggest shutdown port  you able to reach from 10.1.1.1 to 10.1.50.5 and 10.1.50.5 can reach 10.1.1.1

all device post show run again

BB

***** Rate All Helpful Responses *****

How to Ask The Cisco Community for Help

Okay, i shut down (actually i deleted) the vlan 50 interfaces from Core1 and Core2. I thought it wouldn't work because i found out i forgot to make a vlan 50 interface on SRVACC1 - but it worked. i can ping 10.1.1.1.

So now my question is, how come my switch is able to forward a packet from vlan 50 without an SVI? Is it because i configured a an IP address on the interface the windows server is connected to?

Review Cisco Networking for a $25 gift card