Hacker News Reader

Comment by RIMR | Hacker News Reader

[flagged]

> From what I understand, they are just one zero-day away from

3 years vs. a "near peer" level state actor, and it is still standing (well, not now, but you get the point) would beg to differ. In stark contrast to Viasat, which had their terminals bricked at the start of the conflict...

What's the failure mode that would lead to an unrecoverable loss of such a large mesh network?

Any failure that causes a loss of control or communication with the satellites, especially as a result of attack or software update rolled out to many of them that contains a latent bug.

if you could somehow put something into persistent storage on the satellites that causes it to crash every time it tries to start up before it gets to the point where it's able to update the software, then they'd be bricked.

I don't know how Starlink's satellites are architected, but the spacecraft I am familiar with has several different boot images with automatic failover. For that system, you would have to replace/corrupt multiple of the actual boot images to brick the system completely.

what if the boot image itself isn't corrupted but rather there's other persistent data which causes the firmware to get stuck / bootloop?

This is more or less what happened to the Mazda head unit firmware in the link I posted.

There are failsafe images that are fully self contained. I would hope that even Starlink satellites are engineered to a higher standard than an automotive head unit.

There also must be some seriously robust out-of-band systems in place here.

Sounds like you might need to dial back your hyperbole a touch. Seems a tad much on the exaggeration side of things.

Reply on news.ycombinator.com