Sander Vesik writes:

Umm... you mean somebody would seriously consider having a project measured
in millions of dollars and not include trivial small things like SECDED,
memory scrubbing and restarts? You know stuff that is slowly coming even
to low end servers? I would be really shocked...

You shouldn't be; millions of dollars doesn't buy much custom
electronics, let alone all the big hardware, software, and people to
run the program. BTW, we don't know they don't they don't have the
features you mention (except we know they have restarts -- probably
more than 60 of them so far). Plus, they might have other ways of
keeping memory corruption risk low, like very good radiation hardening
or frequent checksumming of memory or something.

I did hear today that they have another copy of the main software
onboard which they can load up if they want to. But I suspect they
greatly suspect some real hardware failure and need to figure out what
that is before trying to work around it. They're even willing to let
the batteries go dead at night while waiting to gather more diagnostic
info, rather than just taking the Microsoft approach and reloading the
software.