Back in 2017 two high-powered GNU/Linux computers were sent into orbit and are still running. They are long overdue for retrieval but are, more than 530 days later, still working. The goal of the project was to test the durability of such systems in preparation for travel to Mars, where data must be processed on site because of the delay in sending it to Earth and then transmitting the results back to Mars. So far autonomous management software has handled all of the hardware problems.
The servers were placed in an airtight box with a radiator that is hooked up to the ISS water-cooling system. Hot air from the computers is guided through the radiator to cool down and than circulated back.
Mr Kasbergen said there had been problems with the redundancy power supply as well as some of the redundant solid-state drives.
But he said the failures were handled by the autonomous management software that was part of the experiment.
The devices will need to be inspected back on Earth to find out what went wrong.
Earlier on SN:
Supercomputer on ISS will soon be Available for Science Experiments (2018)
HPE "Supercomputer" on the ISS Survives for 340 Days and Counting (2018)
HPE Supercomputer to be Sent to the ISS (2017)
(Score: 2) by Runaway1956 on Sunday March 03 2019, @02:14PM (6 children)
6030 days 5 hours 21 minutes 19 seconds on the screen shot here - https://arstechnica.com/information-technology/2013/03/epic-uptime-achievement-can-you-beat-16-years/ [arstechnica.com]
(Score: 3, Funny) by maxwell demon on Sunday March 03 2019, @02:32PM (3 children)
The trick to get long uptime:
Step 1: Boot the system.
Step 2: Suspend to disk.
Step 3: Wait. The longer you wait, the larger your uptime.
Step 4: Bring your system back up.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 0) by Anonymous Coward on Sunday March 03 2019, @04:37PM (2 children)
It's probably possible to get it by fucking with the system clock too... Why anybody would want it I do not know.
I personally install kernel updates that on my OS do require restarting. Of course in an environment like aboard the ISS that's something you don't want.
(Score: 2) by maxwell demon on Sunday March 03 2019, @10:15PM (1 child)
I also don't see why someone would want to fuck with the system clock. On the other hand, there are good reasons why one would want to suspend to disk.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by Freeman on Monday March 04 2019, @04:14PM
You mess with the system clock for games that use it for things like seasonal events or the like. That you don't have to wait until October to get all those cool Halloween Costumes in Terraria.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2) by RS3 on Sunday March 03 2019, @09:49PM
Ahhh, Netware 3.12. Thank you for that link. Brings back really good memories. I built dozens of them in that time period. For all I know some of them still run, but I doubt it because stuff gets depreciated and tossed.
The fact that one had over 6,000 days uptime says it all. Of course that means the hardware had to survive that long, and obviously it was built before the not so great capacitor plague https://en.wikipedia.org/wiki/Capacitor_plague [wikipedia.org].
Netware is so different from the rest it felt like you were doing something on a significantly different and higher level. It was crazy efficient too. I think the elevator seek was a big factor. It would do RAID 0 natively if you were okay living dangerously. Under heavy loads, watching the HD lights and hearing the head actuators moving in that coordinated rhythm was musical. I shipped a couple of them before we started using hardware RAID 5 controllers. Novell did issue a lot of patches over the years- that annoyed me, but it got me ready for the coming onslaught of Windows patches.
BTW, in that article about the 6,000+ day uptime they were worried about the HD's bearings, but I'd bet they were okay. I remember those days well, and still have a couple of full-height 5.25" drives. It's not the bearings- the spindle motor is _very_ noisey. Sounds like something grinding, but it's partly the motor's size and current due to the rotational size / inertial mass, and the likely square wave drive that would create harsh harmonics. Brand new they always sounded like they had cube instead of ball bearings- Maxtor, WD, HP, etc.
(Score: 3, Interesting) by driverless on Sunday March 03 2019, @11:38PM
Also depends on the hardware, for things like Tandem and Stratus systems the uptime is typically "since it was first plugged in". There's a company I know who were still, the last time I heard, running a Tandem CLX. It ran through the Loma Pieta earthquake, among other things.