How One Router Grounded America (Part 1 of 2)

This is part one of a two-part analysis of the FAA national ground stop on Friday, November 13th, 2026. This post covers what has been reported and our best guess at the technical cause. Part two (here) covers the procedures that would have prevented it.

Disclaimer: We have no inside knowledge of the FAA network. The facts below come from public reporting, and the router-level hypotheses are our own reasoning, not confirmed findings.

What Happened

At 3:54 PM Eastern on November 13th, the FAA grounded every flight in the country, something it had done only once before since 9/11. The stop was lifted at 8:15 PM. According to the FAA, operators canceled approximately 12,462 flights that Friday, including 9,082 by major airlines (the FAA notes not all of those were caused by the ground stop). Nextgov reported the disruption affected more than 2 million passengers.

By Saturday the government had a cause. L3Harris, the FAA's telecom contractor for more than twenty years, was doing routine maintenance at an FAA radio site in El Dorado, Arkansas, moving it from analog copper wire to digital fiber, as Duffy described it. According to Duffy, replacing the router there triggered problems with a router at a radar site in Amarillo, Texas that had been replaced weeks earlier, and the Amarillo router started broadcasting and overloading routers at air traffic control centers across the country: more than 1,000 connection attempts every four to five seconds. Communications broke down at 17 of the FAA's 22 en route centers. The FAA says the network stabilized once L3Harris blocked the two locations.

The new El Dorado router was made by Juniper Networks, and an HPE spokesman told the Wall Street Journal the Juniper devices appeared to have been working as intended. Officials said it did not appear to be a cyberattack, and federal officials and Microsoft said there was no indication the stop was related to Microsoft's global outage the day before.

What they have not determined is the fault itself. In Transportation Secretary Sean Duffy's words: "Was it the router, or were they not doing great work for us at L3Harris?"

Which Protocol Behaves Like This?

The FAA Telecommunications Infrastructure (FTI) network is partway through a move from older circuits to IP. FAA and industry documents describe IP services on a private MPLS architecture with OSPF and BGP routing, and L3Harris job postings for the program list IS-IS, LDP (MPLS label distribution) and HSRP as well. Those are the standard protocols of any large carrier network. So which one of them would produce a flood like this? We started with five hypotheses and ranked them by probability. Then we walked through the cutover step by step, as the engineer on site in Arkansas, and the ranking changed.

This is our reconstruction of a typical copper-to-fiber router cutover, not the FAA's actual runbook:

  1. Build the new router's configuration by copying the old router's configuration as a template.
  2. Stage it and test the fiber.
  3. Drain traffic off the old router so the site runs on its redundant path.
  4. Connect the fiber and verify the light levels.
  5. Bring up the control plane one protocol at a time: OSPF first, then BGP, then label distribution, then the gateway redundancy protocol.
  6. Verify the routing table and shift traffic onto the new router.
  7. Watch it through a soak period.
  8. Decommission the old router.

Every step has a gate, and a flood like this one means a gate failed. Here are the top two hypotheses in their new order.

Hypothesis 1: Two Routers, One Identity

Every router carries a unique router ID, and OSPF uses it to tell routers apart. OSPF also spreads updates about the network with a process literally called flooding. Now look at step 1 again: the new router's config is a copy of the old router's config, and the most common cutover mistake is forgetting to change the identity fields in that copy.

If the new router came up with the same router ID as the old one, and the old one was still alive because the drain wasn't complete, you have two routers answering to the same name. Each keeps flooding updates to override the other, and the fight does not stop on its own. The same kind of collision can happen with HSRP, the protocol that decides which router is the active gateway: two routers both claiming the job, traffic flapping between them, sessions resetting.

Three things make this our leading explanation:

  1. The mechanism matches the symptom. Flooding is what OSPF does, and an identity fight never settles by itself.
  2. Two new routers. The union representing FAA safety employees said two routers were updated on Friday. Duffy later said the Amarillo work was done weeks ago, and that the El Dorado replacement triggered the Amarillo router. Either way, two recently replaced routers from the same program misbehaved together. If both were built from the same template, a duplicated identity sits quietly until the second router comes up, and then the fight starts. The collision doesn't have to be new versus old. That explains the timing better than two unrelated failures.
  3. The procedure sets the trap. Cloning a config is how the job is done, and the identity fields are the easiest thing to miss.

Hypothesis 2: A BGP Session Flap Storm

BGP is how routers exchange routes between networks, and it runs on top of TCP. A router that cannot establish a BGP session makes literal TCP connection attempts, over and over. If the new router had a misconfigured peer (wrong autonomous system number, wrong neighbor address, bad authentication password), it would hammer its peers with retries, and every failed session would also send route withdrawals across the backbone.

This was our original number one. It fell to number two on the math. 1,000 attempts every four to five seconds works out to 200 to 250 per second. A single router retrying a single peer makes a few attempts per second, not two hundred. That rate only makes sense as an aggregate: the routers at all 17 affected centers retrying their sessions at once after something underneath them broke.

In my opinion, BGP flapping is probably what the flood looked like, not what started it.

What Would Settle It

The router logs at the two sites. Here is what each one would show:

What the logs show What it points to
A flood of TCP SYNs to port 179 BGP session flapping
A storm of link-state advertisements The OSPF identity fight
A broadcast spike A Layer 2 loop (our third-ranked hypothesis)

Until the FAA publishes its root cause, the first two are the best explanations we have. Reuters' account says the Amarillo router "started broadcasting," which gives the Layer 2 hypothesis more weight than we first gave it. And HPE says its devices appeared to be working as intended, which points at configuration, not hardware. In my opinion the failure was in the act of replacing the old hardware.

The Old Equipment

The routers being replaced are about 25 years old, Duffy said, and were supposed to be replaced in 2006. The FAA spends $1 billion a year maintaining its old copper wire system. The network didn't fail because that old equipment died. It failed because someone finally tried to replace it.

And this has happened before. On November 19, 2009, 129 of the nation's air traffic facilities lost FAA telecommunications after a Harris technician incorrectly configured an FTI router at Los Angeles Center, during an earlier move to a new network, according to a DOT Inspector General report (here). Harris merged with L3 Technologies in 2019 to form L3Harris.

In part two (here), I go through 13 ITIL practices and show which one would have caught each failure.

Need help? Email us at Hello At PatriotConsultingTech.com

Sources and references:

About the author. Joe Stocker is the Founder and Chief Technical Officer of Patriot Consulting, a former Microsoft Security MVP (2020-2026), and author of the book "Securing Microsoft 365." In his spare time Joe volunteers with several Microsoft programs including Microsoft's Defending Democracy (AccountGuard) program, Microsoft Tech for Social Impact, and the Microsoft Software and Systems Academy (MSSA), which serves military service members desiring to transition into the civilian workspace.

Note: The author created this article with assistance from AI. Learn more