HP2620 Routing loop - problem

[jonare]

Hey

Using a HP2620 with latest ArubaOS release to route a internal network 10.16.160.0/22 to 10.16..108.2 witch is a Cisco 2901 router. The switch has 10.16.108.10 ip on the same VLAN as the Cisco router.

When pinging a unit in the 10.16.160.0/22 network that is turned off a routing loop appears. If the unit is on everything seems normal.

Tracert shows:

32.54.20.97
32.54.20.2
10.16.108.10
10.16.108.2
10.16.108.10
10.16.108.2
10.16.108.10
10.16.108.2
10.16.108.10
10.16.108.2

Ping shows:

Pinging 10.16.160.3 with 32 bytes of data:
Reply from 10.16.108.2: TTL expired in transit.
Reply from 10.16.108.2: TTL expired in transit.
Reply from 10.16.108.2: TTL expired in transit.
Reply from 10.16.108.2: TTL expired in transit.

Attached are "sh ip route" from router and switch. (hpforum1.jpg)

I have the same setup againt an another network, but the router is a old 2626 with provision 10.119 firmware. The problem does not appear in that setup. It seems the 2620 is returning the ICMP packet to the Cisco 2901 in its orginal form, which again is returned back to the 2620 switch.

Any suggestions.

----> Followup ---->

Done some testing on the situation. Seems the problem accours in fw: 16_01_0010 +++  for the HP2620.

The switch acts normal on fw: 15_18_0013.

1  <1 ms <1 ms <1 ms 32.54.20.97
2  <1 ms <1 ms <1 ms 32.54.20.2
3  <1 ms 1 ms 1 ms 10.16.108.12
4  10.16.108.10 reports: Destination host unreachable.

Can we call this a bug or is someting changed that I dont understand ?

Regards

[jonare]

Hey Everyone...

Using the 16.x.x.x branch of aruba firmware i causing problems for the organization im working in. When used on switches that do routing you get the loop problem i described. But also WYSE terminals looses connection to everything but the subnet it is in. All packets that are trying to access next hop gets dropped or something else happens. 

When using 15.x.x.x firmware everything is ok. 

[16again]

The loop in topic start might be explanable:
The Cisco has a static route for 10.16.160.0/22 to HP switch.
The network mentioned is directly connected on HP switch.  If no ports are active on that VLAN, VLAN IP interface goes down removing the connected 10.16.160.0/22 route.  Then default route kicks in, sending traffic back to your Cisco 2901.
(on Cisco , use "no autostate" on vlan interface, HP might have similar command)

But "loop" in 2nd post needs way better problem description