I have a small cluster of SRX220's (v 12.3X48-D30.7). They have a single reth0 interface between them that has a couple subinterfaces tagged. The reth terminates into a pair of EX3300's (Node0 -> VC member 0, Node1 -> VC member1). Works greate.
We have two ISP links. ISP-A is an SDWAN box with multiple aggregated DSL links, and ISP-B is a cellular backup. Normally the RPM used to failover between them is pretty trivial. However this one is REALLY trying my patience and I have to be missing something. Because of the nature of the SDWAN box it's possible that it may be offline without really being offline. SO we are pinging out to a remote IP we own. However the problem I am having is that once it detects the primary SDWAN connection is offline it fails over to the Cellular carrier. However the ping test out the SDWAN interface continues to use the cellular next-hop. You can see it in the logs:
Oct 19 18:11:55 18:11:55.788108:CID-1:RT:Doing DESTINATION addr route-lookup
Oct 19 18:11:55 18:11:55.788108:CID-1:RT:flow_ipv4_rt_lkup success 198.97.x.y, iifl 0x0, oifl 0x79
Oct 19 18:11:55 18:11:55.788108:CID-1:RT:Checking in-ifp from .local..0 to reth0.3 for src: 184.108.40.206 in vr_id:0
Oct 19 18:11:55 18:11:55.788108:CID-1:RT: routed (x_dst_ip 198.97.x.y) from junos-host (.local..0 in 0) to reth0.4, Next-hop: 192.168.0.1
Oct 19 18:11:55 18:11:55.788108:CID-1:RT:flow_first_policy_search: policy search from zone junos-host-> zone Public (0x0,0x6697,0x6697)