HP53xx and Cisco 6500 ip routing

[Kelly Stadel_1]

I have not been able to find enough info in the manual on the 53xx to figure this out and

I've exhausted my searches on the net. Need help in configuring this scenario....

IPs are examples only (all 10.1.x.x with /30 subnet)

10.1 -C6500- 20.1 30.1 -C6500- 40.1
| \ / |
| \ / |
| /\ |
| / \ |
| / \ |
10.2 -HP5372- 30.2 20.2-HP5372-40.2
HP L2 ---VLAN50-----HP L2

Each connection above is directly connected, right now configured in 4 individual VLANs,

VLAN 10, VLAN 20, VLAN 30 and VLAN 40. The HPs are connected at L2 at VLAN50.

We want to connect the HPs to 2 different Cisco's for redundancy at L3. The routing info

and how it relates to the L2 VLANs in the manual is very vague.

So far we've been able to get comms between HPs and Cisco by configuring each direct

connection as different VLANS with each their own /30 subnet. Comms are happening there

however, we still can't ping L2 devices on the HPs or the L2 IP itself.

Problems we are experiencing are:
1. L2 is not communicating with L3 on the HPs or any further into the network.
2. Don't know whats the best to use at L3 between HP and Cisco. Cisco's are using OSPF with

MD5 authentication which I don't see the HPs support so we have limited options to connect

at L3. Options are static routes redistributed on Cisco's (preferable), rip on HPs

redistributed on Ciscos, or new OSPF area with different authentication with a virtual link

( The HPs are connected to Area 2, not directly to the backbone).

I am trying static routes to begin with and even that we are having problems with. So here

are my questions.

Questions:

1. How does 53xx L2 talk with L3 when the default gateway only works while not doing ip

routing on the switch?
2. Where is the communication link between the HP L2 and L3 configured? Is it part of the

VLAN or does a virtual interface have to be set up for it?
3. If routing is part of a VLAN you set up, can secondary addresses (for HP to Cisco comm

subnet) be set up?
4. What address do you use for the hosts Gateway, secondary address or the primary switch

address?
5. Does a ip protocol have to be used if we only want the 2 routes out of the HPs?
6. Do I have to set up static routes from HPs to Cisco's or default routes?
7. Do the static routes point to the Cisco IP as the gateway or to the HPs IP as the

gateway?
8. If I can setup default routes in lieu of static on the HPs, can more than 1 default route

be configured for HP interfaces to maintain redundancy?
9. Would I then have to configure 2 different gateways for the hosts attached to the HPs.


HP routing is all fairly new to me as is routing in general. So far the 53xx router info is

pretty ambiguous. Info seems to be more forthcoming on the 9000 series routing switches.

If someone could give an example configuration to fit my scenario it would be most

appreciated.

Thanks all.

Kelly

[Kelly Stadel_1]

Thank you for your response. You cleared a lot of things up for me.

We have managed to get most everything working now.

The holdout is for providing the HP5372's a 2nd default route for redundancy. The switches do not allow a second ip route 0.0.0.0/0 nor do they allow ip default-network or ip default-gateway (while routing is active on switch).

Is there any other way to provide a redundant default route out of these switches? Or alternatively is there another way to enter a static route that says "everything else but..."?

Thanks,

Kelly

[Ron Kinner]

Don't put any defaults on the 5372's. Turn on RIP or OSPF between 5372s and the next hop and put the defaults on the last redundant switches/routers before the internet. That way the routing protocol will choose which one of the paths and then if it dies will know to use the other one.

Even if it let you use two static routes it wouldn't use the second one unless the direct link died since it has no other way of telling which is alive.

Ron

[Kelly Stadel_1]

Ok. So we're giving up the static route idea and decided to go to ospf.

Still having problems.

HP's do not talk MD5 authentication so we changed the Cisco to HP interface so that there is no authentication at all on this interface....

Still doesn't work. We have:
1.No authentication on both ends,
2.Same area,
3.same subnet on the connecting interface
3.Same hello and dead times

Cannot ping anymore from the HP to the Cisco. Physical link is up as we can see the MACs in both tables. Both ports are configured the same - 100 full duplex

Has anyone tried connecting these HPs to Cisco's and had it work? If so, do you have a config example of how you did it.

Thanks,

Kelly

[Ron Kinner · Accepted solution]

You can try turning on the debug for OSPF packet. That usually tells you why it's not working.

If you are telneting in you have to do
term mon
first so you can see the debug. I also recommend
no logg cons
and
logg buff which gives you the opportunity to reread the debugs if they come too fast for your terminal.

sh log

to see the log files.

debug ip ospf packet

A good trick is to have a second session open at the same time with
no debug all
already typed in so all you have to do is switch to the second session and hit Enter to cancel the debug. This is useful in case the debug gets out of hand. To turn off term mon it's:

term no mon

There are other ospf debugs which may be useful. Do:

debug ip ospf ?

to see what is available.

You say:

"Cannot ping anymore from the HP to the Cisco. Physical link is up as we can see the MACs in both tables. Both ports are configured the same - 100 full duplex"

This is strange and is probably the real reason your OSPF isn't working. There should not be a routing issue here since you say both the HP and the Cisco are in the same subnet but on the Cisco I would look at:

sh ip route A.B.C.D
where A.B.C.D is the address of the HP on the same subnet. I have seen cases where a bad static route on another router overrode the connected interface and caused the router to send the packet away instead of out on the wire. The quick fix for this was to do:

conf t
ip route A.B.C.D ethernetX
end
wr me

where X is the interface that goes to the HP.

debug ip icmp

will show the pings as they come in from the HP. If you don't see them then check the routing/masking/tagging on the HP. (You don't want tagging unless you have configured the cisco to do dot1q and set up subinterfaces to handle it.)

Ron

[Kelly Stadel_1]

I went through the debugging process as well as looking as OSPF params on both Cisco and HP.

On OSPF hellos, seems the Cisco was hearing HP but HP not hearing Cisco, hello stuck in INIT state. Went to Cisco's troubleshooting guide and they said check physical side.

Did that...and guess what...HP connection to translator was only passing traffic one way :-(

We changed the translator(s) and lo and behold the OSPF started to work between the two (they had already been looked at once before).

One problem remains, and that is XRRP is not yet on the HP's. We need the common gateway for the hosts. I understand that software is in the making so we'll have to wait for that.

Thanks again, Ron, for your responses to my questions. You have been very helpful.

Kelly

[Ron Kinner]

Glad I could help. Don't forget to assign points.

Ron