There are two default routes on configured on proxySG as per the techdocs document
Group 1 -> Default Gateway1 -> interface 0:0->Primary
Group 2-> Default Gateway2 -> interface 1:0->Secondary
When the Gateway1 connected to proxySG is brought down, the traffic still continues to egress out of proxysg interface 0:0 corresponding to Gateway1 and it takes more than a minute for traffic to egress out of proxysg interface 1:0 corresponding to Gateway2. So essentially the proxySG takes more than a minute to determine if the gateway is reachable.
ProxySG with multiple default gateways
SS OS version below 7.3.19.1 (in 7.3 release train)
SG OS version below 7.4.3.1 (in 7.4 release train)
Previously, proxySG would send ARP requests every 90 secs to detect if the gateway is reachable.
Starting 7.3.19.1 and 7.4.3.1, proxySG now sends ARP requests every 10 secs to detect if the gateway is reachable. This, combined with the inbuilt ECMP health checks results in appliance now detecting unreachable gateway in 20-30 secs.
Starting SG OS 7.3.19.1 and SG OS 7.4.3.1 , there have been improvements made with respect to detection of unreachable gateway
For networks that distribute traffic through multiple default gateways, the appliance now detects when a gateway is unreachable in 20-30 seconds. Previously, it could take over a minute to detect an unreachable gateway