I haven't been able to find a discussion of this problem, and SonicWALL support is useless as usual, so I hope someone here can shed some light on why it's happening..
Here's the situation:
Two uplinks from the colo (using HSRP) are plugged into a single Catalyst 2924 (without VLANs) along with the WAN port of a SonicWALL Pro 3060. The SonicWALL is bridging, so the HSRP IP, the SonicWALL IP, and the servers on the LAN side are all in a single subnet. The Catalyst does not have an IP address.
The servers on the LAN side have the HSRP IP as their gateway. When both uplinks are connected, the primary advertises the HSRP IP, all of the LAN servers can ping it, but some of the LAN servers are unable to reach or be reached from the Internet. Apparently some of the traffic is trying to leave by the standby HSRP uplink, but the gateway IP is ARPing with the correct MAC of the primary uplink.
Everything works fine when only one of the uplinks is plugged in. Any ideas about the source of the problem or how to work around it?
Strangely enough this never occurred with the SonicWALL Pro 1260, but it happens every time with the Pro 3060 (both on current versions of SonicOS Enhanced).