AT&T Fiber Outage (ONT / OPNsense WAN Loss)¶
Summary¶
Troubleshooting a sudden internet outage in a homelab environment using OPNsense with AT&T fiber. The issue was isolated to the upstream fiber connection (ONT) and resolved via power cycling.
Environment¶
- Router/Firewall: OPNsense
- ISP: AT&T Fiber
- Gateway: BGW320
- ONT (Optical Network Terminal): External fiber termination device
- LAN: Homelab network (Proxmox cluster, Docker workloads, etc.)
- Topology:
- Fiber → ONT → AT&T Gateway (BGW320 or passthrough) → OPNsense → LAN
Problem¶
Complete loss of internet connectivity across the homelab network.
Symptoms¶
- No internet access from LAN devices
- Likely failed external connectivity (e.g., ping to public IPs)
- ONT displayed a red light prior to recovery
- AT&T gateway not in normal solid white state during issue
Actions Taken¶
- Observed physical device status (ONT and AT&T gateway LEDs)
- Performed a power cycle of the ONT
- Allowed ONT to fully reinitialize and re-establish fiber link
- Verified gateway returned to normal (solid white) state
Key Findings¶
- Red light on ONT indicated Loss of Signal (LOS) or upstream fiber issue
- Issue was not caused by:
- OPNsense configuration
- Internal networking (LAN, VLANs, firewall rules)
- Likely causes:
- Temporary ISP-side disruption
- Optical signal drop or PON re-sync requirement
- AT&T fiber can exhibit MAC/session stickiness, requiring ONT reset to renegotiate connection
Resolution¶
- Power cycling the ONT restored fiber signal and connectivity
- Network returned to normal operation without additional configuration changes
Validation¶
- AT&T gateway returned to solid white (healthy state)
- Internet connectivity restored across LAN devices
- Implicit validation: upstream routing, DNS, and WAN connectivity functional
Follow-Up Tasks¶
- Place ONT and gateway on a UPS to prevent transient outages
- Document WAN interface name on OPNsense for faster troubleshooting
- Capture logs during future outages:
/var/log/system.log- Consider monitoring:
- WAN gateway status (dpinger)
- Interface flaps
- (Optional) Review AT&T passthrough / bypass configuration for stability
Lessons Learned¶
- Physical layer issues can mimic higher-layer failures (DNS, routing, firewall)
- Always check ONT status lights first in fiber setups
- Power cycling ONT is often required due to:
- PON re-authentication
- ISP MAC/session binding behavior
- LED meanings differ:
- BGW320: white = healthy
- ONT: green = healthy, red = signal failure
- Avoid unnecessary troubleshooting of OPNsense until WAN link is confirmed up
Command Reference¶
Command¶
ping 1.1.1.1
What it does¶
Tests basic IP connectivity to an external host (Cloudflare DNS).
Why it was used¶
To determine whether the issue was: - Layer 3 connectivity (WAN down), or - DNS-related
Expected Result¶
- Success → WAN is up, issue likely DNS
- Failure → WAN or upstream connectivity issue
Command¶
ifconfig -a
What it does¶
Displays all network interfaces and their status.
Why it was relevant¶
Used to identify the WAN interface and check: - Link state - Assigned IP address
Expected Result¶
- WAN interface should show:
status: active- Valid public IP
Command¶
netstat -rn4 | grep ^default
What it does¶
Displays the IPv4 routing table default gateway.
Why it was relevant¶
Confirms whether a valid default route exists for outbound traffic.
Expected Result¶
- Valid default gateway via WAN interface
Command¶
/sbin/dhclient -r <wanif>
/sbin/dhclient <wanif>
What it does¶
Releases and renews DHCP lease on the WAN interface.
Why it was suggested¶
To recover WAN connectivity without rebooting hardware.
Important Notes¶
<wanif>= actual WAN interface (e.g.,igb0,ix0)- Safe to run
Expected Result¶
- New IP lease assigned
- WAN connectivity restored
Command¶
clog /var/log/system.log | tail -50
What it does¶
Displays recent system logs on OPNsense.
Why it was relevant¶
To check: - DHCP failures - Gateway status - Interface errors
Expected Result¶
- Successful DHCP lease messages
- No persistent WAN errors
Command¶
configctl interface reconfigure wan
What it does¶
Reinitializes the WAN interface in OPNsense.
Why it was suggested¶
To recover from gateway or dpinger issues without rebooting.
Expected Result¶
- WAN interface resets
- Gateway status restored
Command¶
configctl unbound restart
What it does¶
Restarts the Unbound DNS resolver.
Why it was suggested¶
Used when: - IP connectivity works - DNS resolution fails
Expected Result¶
- DNS resolution restored
Command¶
drill cloudflare.com @1.1.1.1
What it does¶
Performs a DNS lookup using a specific DNS server.
Why it was suggested¶
To isolate DNS issues: - Tests external DNS directly
Expected Result¶
- Valid DNS response → DNS working
- Failure → DNS or upstream issue