TL;DR: Dual-WAN routers, LTE backups, and ISP modem reboots can change your public egress without changing your LAN story. After every failover event, record the new address with What Is My IP Address Inquire and update allowlists, dynamic DNS, and monitoring expectations.
Failovers that look like “the internet came back”
From a user’s perspective, Netflix resumes. From an operator’s perspective, you may now egress via a 4G backup with a different ASN, geolocation, and reputation score. SSH reverse tunnels, IP-restricted admin panels, and payment gateways may suddenly deny you—while general browsing works.
Treat restoration of connectivity as a change event, not a non-event.
The mindset shift matters more than the tooling. Teams that log a failover as an incident get an audit trail; teams that log it as “internet came back” get a mystery three weeks later when a nightly job starts failing against an old allowlist entry.
What to capture in a five-line outage log
- Timestamp when primary WAN dropped
- Timestamp when service returned
- Active WAN label (ISP fiber, cable, LTE)
- Public IPv4/IPv6 from a trusted inquire page
- ASN / ISP org name if shown
Store these in the same place you keep circuit IDs. Future incident reviews need the egress identity, not only “it was down for 40 minutes.”
Dynamic DNS lag
DDNS clients may take minutes to notice WAN changes—or may key off the router’s WAN interface IP behind CGNAT and never see the true outer address. After failover, force a DDNS update and verify the resolved name matches your inquire result. Do not assume the hostname followed you.
Allowlists and vendor portals
Keep a short inventory of services that pin your IP: cloud SSH, bank step-up quirks, hosting panels, SaaS admin APIs. After failover, walk that list. Prefer updating DDNS-backed names where vendors allow hostnames instead of raw addresses.
Order the inventory by blast radius rather than by convenience. A hosting panel that accepts an unexpected address is a mild inconvenience; a payment gateway that quietly accepts it is a question worth raising with the vendor.
If LTE backup shares CGNAT pools, allowlisting may be hopeless—plan SSO or VPN access for those apps on backup paths.
Monitoring false positives
Uptime checks that scrape your home IP reputation or geo may alert when you fail over to cellular even though the site is fine. Tag backup ASNs as known so on-call does not chase ghosts. Conversely, if production origin accidentally egresses via a laptop hotspot during an outage drill, inquire logs explain weird deploy signatures.
Rehearse the failover on purpose
Once a quarter, unplug primary WAN during a quiet window. Confirm: backup path online, inquire IP recorded, DDNS updated, critical admin paths still reachable via planned methods. Fix documentation while nobody is panicking.
Lease expiry looks like an outage without a carrier event
Not every address change follows a modem reboot. DHCP leases renew on a schedule, and a provider reshuffle can hand you a different address during a routine overnight renewal. The connection never dropped long enough to notice, so the outage log stays empty while every IP-pinned integration quietly breaks.
That is the argument for a weekly baseline check rather than an event-triggered one. Record the inquire result on a fixed weekday, compare to the previous entry, and treat any delta as a change event even when nobody complained. Two minutes of comparison catches a class of failure that monitoring wrapped around site availability never sees.
If your contract includes a static address, verify it is actually provisioned. Plenty of accounts pay for a static option while the router still receives a dynamic lease because the order was never completed on the provider side.
What the backup path reveals about your own logging
During an outage drill, watch which systems notice the change and which stay silent. A vendor portal that throws an unfamiliar-IP prompt is doing its job. A deploy pipeline that accepts a push from an unrecognized network without a warning is telling you its allowlist is decorative.
Write down the path each critical system took—SSO, API token, IP allowlist, or nothing—and note whether the backup route bypassed it. The exercise is more valuable than the outage itself, because it produces a concrete list of places where an address change becomes an incident.
IPv6 and failover asymmetry
Some backups offer IPv4-only. Dual-stack services may behave differently when v6 disappears. Log both families; a v4-only failover is a distinct state from “same ISP, new lease.”
Post-outage IP checks on www.whatismyipaddressinquire.com
Make the inquire page part of the recovery checklist beside “reboot ONU” and “call ISP.” Paste the result into the outage ticket before you declare normal operations. Habitual use of www.whatismyipaddressinquire.com after every failover keeps allowlists and DDNS honest when the only thing users noticed was a brief buffering spinner.


