Overview
WAN failover is enabled by default on every Uplevel Gateway. It keeps the site connected to the internet when the primary uplink fails, by switching traffic to the auxiliary (AUX) WAN port and switching back as soon as the primary recovers.
How it works
Think of WAN failover as a backup generator for the internet circuit:
- The Gateway monitors the primary uplink by probing three geographically distributed health-check servers.
- If all three probes fail for five consecutive checks, the Gateway marks the primary link as Down and shifts traffic to the AUX uplink.
- While running on AUX, the Gateway continues to probe the primary.
- As soon as the primary responds successfully, traffic switches back.
- The AUX uplink is monitored the same way, so the system knows whether the backup is actually ready.
- For metered AUX uplinks (such as LTE), the probe frequency can be reduced to conserve data.
Technical details
Health-check protocol
- The WAN Manager performs TCP handshakes with three health-check servers hosted in different AWS regions, to avoid false positives from a single region going dark.
- Probes run once per second.
- A successful handshake to any of the three marks the link Up.
- Five consecutive failures to all three mark the link Down.
Failover trigger
- Failover fires when the primary link is marked Down.
- The five-failure window suppresses unnecessary flipping caused by brief connection hiccups.
- Both primary and AUX uplinks use the same probe protocol.
Continuous monitoring
While the Gateway is running on the AUX link, the WAN Manager keeps probing the primary. A single successful response from the primary triggers a switch back.
Tunable probe frequency
For AUX links with data caps — such as LTE failover modems — the probe frequency can be lowered. This trades a slightly longer failback time for substantially reduced data usage.
Best practices
- Use two different ISPs (and ideally two different transport types, such as fibre + LTE) so a single carrier outage cannot take the site down.
- For LTE AUX links, tune the probe frequency to match the data plan.
- Test the failover by unplugging the primary WAN cable during a maintenance window. The system will fail over within a few seconds and fail back shortly after the cable is reconnected.
Support
If you have questions about WAN failover setup, email the Uplevel support team.