3750 trunked to 6509, but cant ping

killa

Hi Guys

I have a somewhat a similar scenerio. I have 2 6509s running hsrp to 10.1.100.254. I have trunked one of the 6509 (with vlan 10 ip address 10.1.100.252) to a new 3750 with Ethercahnnel. I cant ping 6509 from 3750 and vice versa. My etherchannel is up. My 6509 uses vlan 10 as management vlan. I'm running transparent vtp mode on all switches. The new 3750 will have servers on 10.1.60.0/24. I therefore created vlan 10 on 3750, assigned ip address 10.1.100.45/24 and vlan 241 with IP address 10.1.60.254. I have allowed most vlans on vlan 10. (I didint specify the 'switchport trunk native vlan 10' though). I created a default route on 3750 as 0.0.0.0 0.0.0.0 10.1.100.254 (the hsrp used by the 6509s). 6509 is running ospf. I put network 10.1.60.0 0.0.0.255 area 0. If I show 'sh ip route 10.1.60.0' on 6509, there is no match. Do I need to specify a static route 'ip route 10.1.60.0 255.255.255.0 10.1.100.45' in 6509 to route to vlan241 in 3750? I expected that since 10.1.100.252 (vlan 10 ip addr in 6509) is in same subnet as 10.1.100.45 (vlan 10 in 3750) I should be able to ping each address from the 2 switches. Where could be the problem? I DONT WANT TO USE OSPF AS WE ARE PHASING THOIS OUT OF OUR NETWORK.

3750 Config is.............
sh run
Building configuration...

Current configuration : 4688 bytes
!
version 12.2
no service pad
service timestamps debug uptime
service timestamps log uptime
service password-encryption
!
hostname BSSSW41
!
enable secret 5 $1$r1EX$pLPfVQFi4XIBeCCUy86S6/
enable password 7 1042591E54141307
!
no aaa new-model
switch 1 provision ws-c3750g-24t
vtp domain bigone.com
vtp mode transparent
ip subnet-zero
ip routing
no ip domain-lookup
ip domain-name bigdomain.com
ip name-server 2.16.1.x
ip name-server 2.16.1.y
ip name-server 10.1.11.z
!
!
!
!
no file verify auto
spanning-tree mode pvst
spanning-tree extend system-id
!
vlan internal allocation policy ascending
!
vlan 10
!
vlan 241
name HO-LAN-Servers
!
!
interface Port-channel1
description ***Channel to SW12***
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 1-301,306-402,405-4094
switchport mode trunk
!
interface GigabitEthernet1/0/1
switchport access vlan 241
switchport mode access
duplex full
speed 1000
spanning-tree portfast
!

***removed config so that it's not long****

!
interface GigabitEthernet1/0/23
description *** Link to SW12 ***
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 1-301,306-402,405-4094
switchport mode trunk
duplex full
speed 1000
channel-group 1 mode on
spanning-tree portfast
!
interface GigabitEthernet1/0/24
description *** Link to SW12 ***
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 1-301,306-402,405-4094
switchport mode trunk
duplex full
speed 1000
channel-group 1 mode on
spanning-tree portfast
!
interface Vlan1
no ip address
!
interface Vlan10
description ***Management Vlan***
ip address 10.1.100.45 255.255.255.0

!
interface Vlan241
description ***Intel Servers***
ip address 10.1.60.254 255.255.255.0
!
ip classless
ip route 0.0.0.0 0.0.0.0 10.1.100.254
ip http server
!
!
!
control-plane
!
!
line con 0
exec-timeout 0 0
line vty 0 4
password 7 1215551043080D08
no login
line vty 5 15
password 7 1215551043080D08
no login
!
!
end

BSSSW41#exit

networker050184

If there is no route in the routing table to reach the destination then you are not going to be able to ping it. Are you replacing ospf with another routing protocol? It would be a lot easier to use a dynamic routing protocol rather than using static routes.

ITdude

networker050184 wrote:

Are you replacing ospf with another routing protocol? It would be a lot easier to use a dynamic routing protocol rather than using static routes.

I'll second that!

killa

I agree with you guys. The ospf would route the 10.1.60.0/24 network. I expect the 10.1.100.252 and 10.1.100.45 adresses which are in the vlan 10 on both both switches to ping each other as they are directly connected networks. They are not. Once they can ping, I can workout the other bits. The reason I dont want to run a dynamic protocol on the 3750 is because the only subnet there is the 10.1.60.0/24 and if I use the default route there pointing to the 6509, which does all the routing, then I save on precious CPU power as well as bandwidth on the etherchannel cause I wont be passing any routing updates. If you can help me figure out why the 10.1.100.252 and 10.1.100.45 addresses cant ping, I would appreciate. Do I need the 'switchport trunk native vlan 10' on the ports forming the etherchannel or the fact that I allowed almost all vlans on the etherchannel suffice? Thanks for taking time to read!

dtlokee

killa wrote:

I agree with you guys. The ospf would route the 10.1.60.0/24 network. I expect the 10.1.100.252 and 10.1.100.45 adresses which are in the vlan 10 on both both switches to ping each other as they are directly connected networks. They are not. Once they can ping, I can workout the other bits. The reason I dont want to run a dynamic protocol on the 3750 is because the only subnet there is the 10.1.60.0/24 and if I use the default route there pointing to the 6509, which does all the routing, then I save on precious CPU power as well as bandwidth on the etherchannel cause I wont be passing any routing updates. If you can help me figure out why the 10.1.100.252 and 10.1.100.45 addresses cant ping, I would appreciate. Do I need the 'switchport trunk native vlan 10' on the ports forming the etherchannel or the fact that I allowed almost all vlans on the etherchannel suffice? Thanks for taking time to read!

Don't take this the wrong way, but I never thought I would see somone worry about OSPF overhead on a gigabit ethernet link. The problem with the "only one subnet" approach is the future scalability of the solution. I have seen engineers use static routes and then later need to do 3X the work to implement a dynamic routing protocol when the static routes become too much of a pain to manage.

You may have a unidirectional routing issue, does the 6509 have a path back to the 3750 for subnet 10.1.60.0/24?

networker050184

You can use a static route both ways, but like dtlokee stated for future growth you would need to add more static routes each time you add more or change the size of your subnets. Trust me all networks grow or change eventually. Also the bandwidth that is used by the ospf proccess would probably not be enough to out wiegh the advantages of dynamic routing.

If you can help me figure out why the 10.1.100.252 and 10.1.100.45 addresses cant ping, I would appreciate.

Like I stated in my first post, the 6509 doesn't have a route back. Once the ping reaches the 6509 it has no way to return it to the 3750. Sends it straight to the bit bucket or the default route (which I highly doubt is pointed back to the 3750).

killa

I fixed it!! Thanks guys. The reason I couldnt ping between the 2 ip addr in same subnet is because my 6509 had mtu 9216 on the Po41 and my gi ports had 1500. BUT I WANT you to read futher because there is some history to it which one of will bump into sometime. Initially my 6509 was configured with some Po on mtu 1500. Like you said network grows. Then at some point there was a need for jumbo frames and some gi were configured for mtu 9216. Since then when you create a Po, it assumes mtu 9216. If the gi's are on 1500, the Po shows mtu 1500 when you type 'sh int po XY'. However if you check log 'sh log', y'll see that the reason for gi ports giving inconclusive results is cos the Po is at 9216 (contrary to what 'sh int po' says. when i first wrote to you guys, the po's were showing up and the gi up as well, but couldnt ping 2 ip addr in same subnet. So what i did was to put 9216 on 6509 po and gi's. On the 3750, I'm just running the 1500. I also added a static to other subnet. (as i said this is temp. we have a plan to run eigrp throughout, so a dynamic routing prot will be in place soon). Thanks for this.

ANOTHER test I did that gave me some wierd results was I trunked my 3750 to 3650 and used a straight cable with mdix auto and link failed. Plugged in cross cable and it worked. On my 6509 to 3750, I use straight cables with mdix auto and it works. So if you happen to us auto somewhere, treat it with a pinch of salt.