Telstra CEO ‘deeply sorry’ for outage and admits risk of time-keeping failure was known

https://www.theguardian.com/business/2026/jul/10/telstra-ceo-deeply-sorry-for-outage-and-admits-risk-of-time-keeping-failure-was-known-ntwnfb

21 points · 4 comments · view on lemmy.world

4 Comments

pHr34kY@lemmy.world · 12 pts · 44d (1 reply)

The significance of the date was that the system used a week counter up to 1,023 weeks, and the default date went back 19.6 years – as with the Telstra issue.

This is a very well documented thing that anyone running a time server should be well aware of.

https://en.wikipedia.org/wiki/GPS_week_number_rollover

What's strange is that the last rollover was in 2019. How did it go this long before failing?

In 2006, the government changed the daylight saving start date to work better during the Commonwealth Games. Telstra misses the memo, and the clock on every phone in Victoria (and possibly NSW) was out by an hour. Their solution? Wait a week unti DST lined up with their unpatched system.

Telstra must really, really hate preventative maintenance work.

SteveTech@aussie.zone · 4 pts · 44d

What's strange is that the last rollover was in 2019. How did it go this long before failing?

It probably wasn't the week counter rolling over, but rather it forgetting that it had rolled over in 2019.

They previously said this was caused by a software update, so I guess that update may have reset that rollover counter.

timmytbt@sh.itjust.works · 2 pts · 44d

Not sorry enough tbh

Salvo@aussie.zone · 1 pts · 39d

Jeff Gerling had been researching and experimenting with time servers recently. He has a brief explanation on the best theory on what happened.

https://youtu.be/1T9xQy-dsQo