IG: Last Major CBP System Outage Caused By Code Defect Known for 2 Years

A passenger processing system outage in August 2019—back when leaving the house, let alone the country was a normal thing—that left arriving travelers stuck for hours waiting in line at customs at airports across the nation could have been avoided with better patch management, according to an inspector general review.

The Customs and Border Protection agency relies on several databases to screen travelers entering the U.S., one of which crashed on Aug. 16, 2019. While the outage itself was avoidable, a new report from the Homeland Security Department Inspector General shows the agency also failed to properly train employees on backup procedures, exacerbating the problem.

The August 2019 outage was not the first—in fact, it was the third such outage in three years. While the IG report cites a January 2017 incident, CBP’s screening system also went offline for several hours at the start of January 2018, leaving travelers stranded on New Year’s Day.

However, after the January 2017 outage, the inspector general began looking at the system and made several recommendations to improve the agency’s response time in the event of future outages. CBP implemented those fixes but they did little to help when the August 2019 outage hit, according to the latest report.

“We conducted this review to determine why CBP’s actions to implement previous OIG recommendations did not prevent the onset and length of the August 16, 2019 nationwide outage,” the report states.

The August 2019 incident stalled passenger processing for more than two and a half hours, causing long lines and delays at major airports across the country.

“According to CBP, the systems were running again by early evening on the East Coast, and there was ‘no indication of any nefarious activity’ during the outage,” the IG report states. “CBP officers were also still able to access security-related databases, though not easily, and maintain security standards while screening people manually.”

But the entire situation could have been avoided with better configuration and patch management, according to the IG.

“We determined CBP’s critical passenger applications were operating on an Oracle database device that was not properly configured and did not have up-to-date patches,” the report states. “This lapse occurred because the Oracle patch did not execute properly and CBP did not take steps to ensure its configuration management policies and procedures were followed and patches were applied promptly.”

The agency could have avoided the outage if IT officials had verified that the database patches were applied properly and in a timely manner, the report states, as Oracle identified the code issue in October 2017.

CBP officials told the IG all but one of the 18 systems using the buggy code had been patched in late 2017 and early 2018, though the oversight office could not confirm that “due to a lack of CBP documentation.” The IG was able to confirm that a key database—Data Machine DM31—was not properly patched when the August 2019 outage occurred.

The outage led to long delays in the arrival screening process as CBP officers had “to revert to less effective backup systems.”

Even so, officers weren’t always able to take advantage of those offline backup systems, as many weren’t trained in how to access and use those systems and received “ineffective communication from CBP headquarters during the outage,” the IG said.

But implementing previous recommendations did make the system more resilient, enabling CBP officials to identify and fix the issues faster than in the past.

The report cites four past recommendations concerning system availability, performance metrics and alert and recovery policies as helping in the most recent outage, though past reports did not cover configuration and patching concerns.

As with past reports, the IG recommended improving “training, procedures, processes and employee awareness to mitigate the risks posed by any future system outage.” CBP officials agreed with all five recommendations made in the latest report.