Hosts using 2900 as gateway stop responding

[Daniel Friedhoff]

Short: connected via VPN, hosts on my nets using either 2900 as a gateway are unreachable - sometimes.

Long: Sonicwall Pro 2040 is hub VPN device. Remote offices (5 of them) are connected as spokes. Each remote office has one subnet for LAN and one subnet for WLAN - both nets of each office are accessible from hub.

2040 is connected to 2900-24, and 2900-24 to 2900-48 via A2/A3 trunk. Both 2900's route, and have their own group of VLANs. Those all work great. From the office, everything works as I would like.

10.0.0.0/24 is LAN on 2040. 10.10.0.0/19 are VLANs on 2900 switches. 10.0.0.0/24 uses 2040 address as gateway, all other nets use either of the two 2900's as gateway.

When I connect via VPN client, I get an address in the 10.0.0.0/24 net (correct). Initially, all hosts on all nets respond. Approx. 2 minutes after connecting, all hots using either 2900 as gateway stop responding. My connection to VPN endpoint is solid. Approx. two minutes later, hosts using 2900's as gateway start responding. Two minutes later they stop, two minutes after that they start, etc. During this time, all remote nets from spoke offices are available. Also, some devices on the 10.0.0.0/24 net use the 2040 as gateway, and those are reachable when everything else is down.

I've been all through the Sonicwall, and with the help of their support group, they feel the device is configured properly.

I'm beginning to feel this is switch related, since anything using the 2900's as a gateway is unavailable to me.

Does the approx. two minute time ring a bell for anyone? What should I check to verify my switch is configured OK?

Thanks.

[Pieter 't Hart]

both switches point to each other as the gateway for 10.10.0.0/16.
this does not seem correct to me.

hostname "MDF-2900-24G"
vlan 1
ip address 10.0.0.2 255.255.255.0
ip route 10.10.0.0 255.255.0.0 10.0.0.3

hostname "MDF-2900-48G"
vlan 1
ip address 10.0.0.3 255.255.255.0
ip route 10.10.0.0 255.255.0.0 10.0.0.2