Incident ResponseOpinion

When the Fix Is the Outage: What USDA's Password Reset Teaches About Credential Rotation

November 22, 2026 · Brian Deitch

Yesterday I wrote that outages that aren't attacks are still a security problem. Today's topic is the flip side: sometimes the security response is the outage.

What We Know About USDA

Starting Monday, November 16, USDA systems that farmers and staff rely on for payments and loans started going offline. Here's what's actually on the record, as reported by Nextgov/FCW (via BNO News), ABC News and others:

  • An internal email from the acting assistant CIO for the Farm Production and Conservation Business Center said the department's cybersecurity team "uncovered a security incident," and staff had to reset database passwords across environments. Applications that depended on those credentials had to be restarted and went offline.
  • A USDA technical bulletin warned apps could have intermittent outages throughout the week.
  • Undersecretary Richard Fordyce, quoted via DTN, called it a "cyberthreat" and said on Friday that most systems should be back online by close of business.
  • ABC reported "multiple sources" calling it a cyberattack. USDA has not publicly confirmed a breach, named an attacker, or said whether any data was taken.

So I'm not going to speculate about who or how. The lesson I care about is already visible: when the department had to rotate credentials, the rotation itself took payment systems down for days.

Rotation Is a Resilience Test You Will Eventually Take

Every incident response plan says "rotate credentials." Very few organizations have ever rotated all of them at once. When you finally do it under pressure, you find out:

  • Which apps have database passwords hard-coded in config files nobody owns.
  • Which service accounts are shared across ten systems, so one reset breaks all ten.
  • Which apps need a restart (or a redeploy) to pick up a new secret.
  • Which vendor-managed systems need a ticket to the vendor before they'll work again.

None of that is unique to government. I've seen the same pattern after ransomware, after a leaked key on GitHub, and after an employee departure that should have been routine. The organizations that recover in hours, not days, are the ones that practiced.

What To Do This Quarter

  • Inventory your secrets and where they live. Every database credential, API key and service account, plus the apps that use it.
  • Kill shared service accounts. One credential per app makes rotation surgical instead of a blackout.
  • Move secrets to a vault and have apps fetch them at runtime, so rotation doesn't need a code change.
  • Prefer short-lived credentials (workload identity, managed identities, token-based database auth) so there's less to rotate in a hurry.
  • Run a rotation drill. Pick one non-critical system, rotate everything it touches, and time the recovery. Then do a bigger one.
  • Write the order of operations down. Which secrets rotate first, which apps restart in what order, who calls which vendor.
  • Pair it with access you can cut quickly. If users and apps reach systems through brokered, identity-based access rather than flat network paths, you can contain an intrusion by revoking access instead of shutting everything off. (Disclosure: I work at Zscaler, which sells that kind of access. The principle holds no matter whose product you use.)

Bottom Line

USDA did the right thing by resetting credentials when it found an incident. The pain came from how hard that reset was. Make rotation boring before you need it, because the day you need it is the worst possible day to learn how.

Sources: BNO News on Nextgov/FCW's report, ABC News, Nacionale, Daily Security Review

usdaincident-responsecredential-rotationsecrets-managementresiliencegovernmentoutagesservice-accounts
All postsDiscuss on LinkedIn