Telstra says bad NTP server caused Australian mobile network outage

Telstra told an Australian Senate inquiry that a network time protocol server caused its recent mobile network outage after maintenance work brought the device back online with its clock set to 2006. The incident disrupted mobile connectivity across Australia, including access to the 000 emergency services number, electronic payments and transport networks, according to the carrier.

The disclosure came in Telstra’s submission to a Senate inquiry into emergency services outages. That inquiry had initially focused on an outage at rival Optus, which has been linked to multiple deaths after callers were unable to reach emergency services.

Telstra’s account points to a routine resilience fix that exposed weak controls around a core dependency: time. The company said the work began shortly before midnight on July 7, when a technician started replacing a chassis that housed one of its NTP servers. The reason for the replacement was a faulty backup power feed.

By 3:38 a.m. on July 8, the technician had completed the chassis replacement and powered the server back up. Telstra said the server included a GPS card that failed to behave as expected after restart. The company attributed that behavior to an earlier design change made to the equipment to address a prior fault, which it said had not been adequately documented.

Telstra also said a software update for the equipment had not been applied, even though the company knew it was available in early 2026. In its submission, Telstra said the outage may have been avoided if the update had been completed, or if the previous design change had been properly reviewed, documented and reflected in the maintenance process.

Bad time data broke certificate checks

After restart, the affected NTP server reset its time to 2006 and began distributing that incorrect time to other machines on Telstra’s network. Network equipment that received the wrong timestamp then assessed digital certificates against an invalid clock and rejected connections.

Telstra said only one of its three NTP servers had the problem. That was still enough for incorrect timing data to reach enough systems to cause a broad service failure.

The carrier isolated the faulty server at 7:11 a.m. and said it had identified every network component that had used the incorrect time data by 10:30 a.m. Service recovery was not instant. Telstra said that as correct time data spread back through the network, some equipment did not close IP sessions, which meant some customer devices could not reconnect unless they were restarted.

Telstra called the outage “clearly unacceptable” and said the incident showed its controls were not strong enough if maintenance work could trigger a failure of that scale. The company said its investigation would examine why the design change was not documented, why the software update was not completed and how known risks should be captured and closed before they affect customers.

The company said it would continue to participate in inquiries, cooperate with the relevant regulator’s investigation and take steps intended to prevent a similar incident. Telstra did not disclose in the submission how many customers were affected or how many emergency calls failed during the outage.

This story draws on original reporting from The Register.