INCIDENT REPLAY · 2025-09-05

BART systemwide shutdown after overnight network upgrade

All service was suspended on a Friday morning until about 11:45, affecting roughly 40,000 commuters, after an overnight upgrade to the communications and computer network left the control centre unable to monitor trains.

A reconstruction, not the operator's network. Hammerhead's authors rebuilt the change from the public record as a small network of their own. No configuration here is the operator's. What follows is what the tool reported on that reconstruction; it is not a claim that the tool would have prevented the outage.

What happened

Who: Bay Area Rapid Transit (BART)

The summary at the top is paraphrased from the sources below. Hammerhead's catalog rates its confidence in this account medium.

Sources

What the public record says

All service was suspended on a Friday morning until about 11:45, affecting roughly 40,000 commuters, after an overnight upgrade to the communications and computer network left the control centre unable to monitor trains.
Who: Bay Area Rapid Transit (BART)

What the reconstruction assumed

Decided by Hammerhead's authors to build a network that runs. None of it describes the operator's network.

Everything below is ours.

The family template (tools/replays/gen_family_policy.py): OCC-1 (the operations control centre) reaches FIELD-1 (the wayside field devices) through OCC-CORE-1 (the core upgraded overnight), which filters inbound on its outside interface. The intent file names the flow: the control centre reaches the field devices on tcp/502.
BART has not published what the overnight change was. The replay assumes the upgraded core's filter lost the permit for the control protocol between the control centre and the field devices. The shape is the finding; the line is a guess.
Cisco IOS syntax and stand-in address space throughout. A reconstruction, not any operator's configuration.

What Hammerhead said

Each run below is pinned by Hammerhead's replay tests at commit 0ee9732: the verdict and the quoted text are checked against the tool's output, and a run that disagrees fails the test. Reason lines are copied verbatim; a pinned line can be part of a longer reason. Every preflight run uses --max-blast-radius 100, so a change is never counted as caught merely because it touched many devices of a small network.

the change as intended

now: pass

On the reconstruction, hammerhead preflight on a before and an after snapshot with --policy @bart_2025_09/policy.yaml --allow-flow-changes returned pass.

First run, before the fixes this batch prompted

first: fail

policy gate: high/critical failure(s) on the 'after' network: the control centre reaches the field devices

Pinned in tests/gauntlet/replays-policy.yaml.

the change as pushed

now: fail

On the reconstruction, hammerhead preflight on a before and an after snapshot with --policy @bart_2025_09/policy.yaml --allow-flow-changes returned fail, for the reasons below.

Reason text pinned (some reason contains each line)

policy gate: high/critical failure(s) on the 'after' network: the control centre reaches the field devices
NOT reachable (expected reachable)

First run, before the fixes this batch prompted

first: fail

policy gate: high/critical failure(s) on the 'after' network: the control centre reaches the field devices

Pinned in tests/gauntlet/replays-policy.yaml.

What changed in Hammerhead because of it

From the “What was fixed” table in Hammerhead's replay notes, where each fix landed after a failing test. “Found by” is copied as written; a family name means the batch this incident belongs to.

A flow rule names its protocol and port

Every reachability rule probed TCP/80, so an intent about SIP could not tell the push from the intended change: 7 of 11 intended changes failed on main and the one that dropped only the named port passed. protocol: and dst_port: on the rule; the policy gate's reason now carries the violation, not only the policy's name.

Found by: Required-flow family

Limits

The configurations are a reconstruction. Where the record is silent, the assumptions above fill the gap; if one is wrong, the replay is wrong in that respect.
A verdict describes the reconstructed network's converged state. Timing, traffic load, and anything a device does that no configuration states, such as software reacting to a health check or a firmware bug, are outside it.
A pass is reported as a pass. A fail says a gate fired on the reconstruction, nothing more.
Why the catalog counts this outage as partly in scope: A planned overnight network change that left the control centre cut off from field devices is the pre-push reachability case, but BART has not published what the change was.

Imported from docs/replays/bart-2025-09/FACTS.md and the replay manifests in the Hammerhead repository at commit 0ee9732.