And the current story today is that the flash is actually fine-- it's either a problem with the file system code used to access it (which can be patched), or some transient corruption caused by solar flare damage. It sounds like they're quite optimistic they can get the rover substantially back up, but it will take weeks.
That's definitely planned if they can be sure it's software, which is still the leading theory. They're also planning on using workarounds with Opportunity that will avoid straining the file system overmuch.
Here's the most I've figured so far, from transcripts of the media briefings combined with my own guesses. There are three kinds of memory on the rover: the RAM (ordinary DRAM, sounds like), the flash which is used for non-volatile buffering of data, and some sort of sturdier EPROM-like programmable memory that holds the system firmware. The flash has some sort of file system running it; I don't know if it's some off-the-shelf file system like what you get with a commercial embedded operating system, or something custom-written. I'd guess the former but I could be wrong.
Anyway, managing the file system consumes RAM, probably for cache and for manipulating catalog information and such. They've found that the amount of RAM consumed scales with the number of files on the flash, and after 18 sols in operation they had so many files on there that they started running out of RAM. They hadn't found this in the simulations on Earth because the exercises didn't last that long; the longest one was 9 days.
Even if there's no easy patch, the workaround for this, if it's really the problem, is pretty obvious: just be really fastidious about cleaning up unneeded files. It sounds as if they weren't worrying about that at all. There isn't nearly enough flash on board that they could just have been hoarding all the science data on the rover; I'd guess it was little metadata files and such.
They're also still looking at the possibility of hardware problems with various components, particularly the high-gain antenna motor, but it sounds like this is now considered less likely.
One of the things that makes progress slow is that they're very cautious about the high-gain antenna (which might conceivably be the source of the trouble, and, if nothing else, needs a substantially functioning system to keep it pointed, since it's directional). So they're using the low-gain antenna for communicating with Spirit; that's the thing in the pictures that looks like a vertical metal pipe with rings at the top. The bandwidth is pretty low, and to do a lot, they need both Martian daylight to power the system, and a line of communication to the rover.
As Earth and Mars get further from opposition, the times of a direct line-of-sight coincide less and less with daytime, so they only get that for part of the day. The rest of the time, they communicate with the rover through the three working artificial satellites of Mars: Mars Global Surveyor, Mars Odyssey, and now Mars Express. These are in low orbits that are essentially polar, so they pass near the site a couple of times each day (near sunrise and sunset, I think, since, like Earth mapping satellites, they want to stay close to the terminator for the best terrain relief). They're not high up enough to work as traditional comsats, but they have RAM in them and support a common system for buffering data and commands and retransmitting them at a later time in the orbit. It's not ideal, but it affords a few more opportunities to communicate.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
timeasmymeasure
no subject
Date: 2004-01-26 01:16 am (UTC)I've encountered a bad flash card or two in my time. In fact, I was sent to a site once for the sole purpose of swapping out said bad flash card.
no subject
Date: 2004-01-26 02:28 am (UTC)no subject
Date: 2004-01-26 03:25 am (UTC)no subject
Date: 2004-01-26 03:36 am (UTC)no subject
Date: 2004-01-27 05:26 am (UTC)Roversoft v1.1b?
no subject
Date: 2004-01-27 12:57 pm (UTC)Here's the most I've figured so far, from transcripts of the media briefings combined with my own guesses. There are three kinds of memory on the rover: the RAM (ordinary DRAM, sounds like), the flash which is used for non-volatile buffering of data, and some sort of sturdier EPROM-like programmable memory that holds the system firmware. The flash has some sort of file system running it; I don't know if it's some off-the-shelf file system like what you get with a commercial embedded operating system, or something custom-written. I'd guess the former but I could be wrong.
Anyway, managing the file system consumes RAM, probably for cache and for manipulating catalog information and such. They've found that the amount of RAM consumed scales with the number of files on the flash, and after 18 sols in operation they had so many files on there that they started running out of RAM. They hadn't found this in the simulations on Earth because the exercises didn't last that long; the longest one was 9 days.
Even if there's no easy patch, the workaround for this, if it's really the problem, is pretty obvious: just be really fastidious about cleaning up unneeded files. It sounds as if they weren't worrying about that at all. There isn't nearly enough flash on board that they could just have been hoarding all the science data on the rover; I'd guess it was little metadata files and such.
They're also still looking at the possibility of hardware problems with various components, particularly the high-gain antenna motor, but it sounds like this is now considered less likely.
no subject
Date: 2004-01-27 01:13 pm (UTC)As Earth and Mars get further from opposition, the times of a direct line-of-sight coincide less and less with daytime, so they only get that for part of the day. The rest of the time, they communicate with the rover through the three working artificial satellites of Mars: Mars Global Surveyor, Mars Odyssey, and now Mars Express. These are in low orbits that are essentially polar, so they pass near the site a couple of times each day (near sunrise and sunset, I think, since, like Earth mapping satellites, they want to stay close to the terminator for the best terrain relief). They're not high up enough to work as traditional comsats, but they have RAM in them and support a common system for buffering data and commands and retransmitting them at a later time in the orbit. It's not ideal, but it affords a few more opportunities to communicate.