Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the
current reference desk pages.
August 25 Information
Revision control for OS
Are there any revision control systems aimed at (or directly integrated into) operating systems which would allow to keep track of modifications to the configuration of the OS? I understand that I could manually add configuration files to any revision control system, just wondering whether this has already been done. I'd be particularly interested in (debian) linux, but out of curiosity also interested in any other OS.
bamse (
talk)
08:40, 25 August 2012 (UTC)reply
Not designed for the purpose, but could
Subversion be suitable for the task (at least for keeping track of changes in /etc)? I know some people use Subversion as a backup tool and change control system for their /home directory (see for instance
this thread). AFAIK, Subversion does a decent job with binary files, too
[1]. This is just a thought, I haven't used Subversion for these purposes, and there may be downsides that I'm not aware of. --
NorwegianBluetalk11:56, 26 August 2012 (UTC)reply
I noticed a warning message in the disk utility of my Linux server recently. See screenshot. Is this an indication that I should replace the disk, or is it a more benign warning message? Thanks, --
NorwegianBluetalk15:18, 25 August 2012 (UTC)reply
.
I would certainly want to have everything on that disk backed up properly. If the disk is mission-critical (that is, if you'd suffer significantly from the downtime associated with replacing it and restoring that backup) then I'd replace it now. If not, you may wish to risk keeping using the disk. That number should certainly be zero, but its being nonzero isn't evidence enough to assume impending failure. If the number increases, that's likely evidence the disk is dirty, scratched, or contaminated, and I'd replace it then. --
Finlay McWalterჷTalk15:27, 25 August 2012 (UTC)reply
Here's the meaning of those four numbers at the bottom:
"Normalized": A manufacturer-defined measure of healthiness where 100 (sometimes 200) is normal and smaller values are worse.
"Worst": The lowest normalized value this parameter has ever had on this drive.
"Threshold": The normalized value at which the drive should probably be replaced (according to the manufacturer).
"Value": More detailed information; in this case, the actual count of reallocated sectors.
There was a Google study a while back showing that SMART-reported errors of any sort were highly correlated with impending drive failure. If it were my drive I'd take it out of service immediately, rather than backing it up and waiting for it to fail.
69.228.170.132 (
talk)
02:25, 26 August 2012 (UTC)reply
It's not an error. The health parameter is at 100%. I don't know why this particular SMART monitor treats any nonzero raw value for this parameter as worthy of a message at all. Most wouldn't. There's a green circle next to the message, which I suppose is meant to mean that it's purely informational. --
BenRG (
talk)
04:52, 26 August 2012 (UTC)reply
Thanks everyone! I experienced a similar situation with the same computer a year ago. It then had three disks, which all were new. When I returned from summer holidays, all three disks had SMART-reported error messages similar to the one showed here. I then replaced all three disks. Given the time of year that these problems have occurred, I wonder if it may have been caused by transients on the power grid during thunderstorms. All the circles were green then, too. The message next to the circle when absolutely everything is ok, is "Disk is healthy". I also noticed now, that there is a mouseover message associated with the Reallocated Sector Count warning, that says:
Type: Failure is a sign of imminent failure (Pre-Fail).
Updates: Every time data is collected (Online)
Raw: 0x050000000000
I'm not really sure what to make of that information, i.e. if it is saying that there actually is a failure condition now, or if it is saying what the interpretation would be if there were a failure condition. To be on the safe side, my immediate action will be to switch the roles of the two 1.5 TB disks on the system (the problem disk holds the /home partition, the other 1.5 TB disk mirrors /home, rsynced every night). --
NorwegianBluetalk10:32, 26 August 2012 (UTC)reply
It's saying what the interpretation would be if there were a failure condition (i.e., if the normalized value dropped below 36). This interpretation is also chosen by the drive manufacturer. There's some information
here.
I'm kind of depressed that Google's data is convincing people to send perfectly good drives to the dump out of fear of random failure (most of their drives with a few reallocation events didn't fail). But I guess it's cheaper than RAID mirroring if downtime is expensive but not too expensive. --
BenRG (
talk)
20:04, 26 August 2012 (UTC)reply
From the Google study "Failure Trends in a Large Disk Drive Population": "As with scan errors, the presence of reallocations seems to have a consistent impact on AFR for all age groups (Figure 7), even if slightly less pronounced. Drives with one or more reallocations do fail more often than those with none. The average impact on AFR appears to be between a factor of 3-6x." AFR is Annualised Failure Rate. So it does appear that this is indicating a higher than normal liklihood of disk failure. --
Phil Holmes (
talk)
10:00, 26 August 2012 (UTC)reply
There are many standard methods to communicate to an embedded device. You are undoubtedly familiar with the USB protocol; many modern devices have hardware support for USB, so you can program and debug using USB. Another common standard is
JTAG, which specifies certain low-level communication methods and defines a hardware interface. JTAG is supported on many embedded devices, even when USB is not. Other embedded computers use less common protocols; or they use customized technology created by the manufacturer or designer.
To run a program on an embedded device, you can either boot directly to the program; or you can run a (simple)
operating system on the device. To boot directly, your program must know intricate details of the hardware, and manage the low-level device bring-up. For this reason, many embedded systems come with a ready-to-use operating system. The operating system will specify how to load and run external programs. Many new systems use full-fledged file systems that are compatible with desktop computers. Often, an embedded computer can attach an
SD card and treat it as a "hard disk" with a standard file system; the embedded operating system loads the program file and executes its main function. Other embedded devices can be programmed over JTAG, and the program may reside in RAM only until the next reboot; or can be stored to a nonvolatile memory.
Nimur (
talk)
17:39, 26 August 2012 (UTC)reply
Legality of redistributing software available for free download
What is the legality of redistributing software that's available for free download, without permission from the copyright holder? Specifically, several companies make available software programs as a large (>2 GB) DVD images for free download. I would like to republish the ISOs on my web site as BitTorrent links, to help those with unstable Internet connections or unstable computers. I understand that copyright prohibits copying and redistribution, but does it make a difference that the copyright holder is not trying to copy-restrict the ISOs themselves, instead requiring an activation key to use the software?
63.152.89.215 (
talk)
21:01, 25 August 2012 (UTC)reply
Check the license distributed with the software. It will tell you whether redistribution is allowed. Just because something is provided for no charge (gratis) does not mean it is free for you to redistribute — it does not mean that copyright does not apply, it does not mean that your own redistribution of it would not be a copyright violation. --
Mr.98 (
talk)
23:08, 25 August 2012 (UTC)reply
Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the
current reference desk pages.
August 25 Information
Revision control for OS
Are there any revision control systems aimed at (or directly integrated into) operating systems which would allow to keep track of modifications to the configuration of the OS? I understand that I could manually add configuration files to any revision control system, just wondering whether this has already been done. I'd be particularly interested in (debian) linux, but out of curiosity also interested in any other OS.
bamse (
talk)
08:40, 25 August 2012 (UTC)reply
Not designed for the purpose, but could
Subversion be suitable for the task (at least for keeping track of changes in /etc)? I know some people use Subversion as a backup tool and change control system for their /home directory (see for instance
this thread). AFAIK, Subversion does a decent job with binary files, too
[1]. This is just a thought, I haven't used Subversion for these purposes, and there may be downsides that I'm not aware of. --
NorwegianBluetalk11:56, 26 August 2012 (UTC)reply
I noticed a warning message in the disk utility of my Linux server recently. See screenshot. Is this an indication that I should replace the disk, or is it a more benign warning message? Thanks, --
NorwegianBluetalk15:18, 25 August 2012 (UTC)reply
.
I would certainly want to have everything on that disk backed up properly. If the disk is mission-critical (that is, if you'd suffer significantly from the downtime associated with replacing it and restoring that backup) then I'd replace it now. If not, you may wish to risk keeping using the disk. That number should certainly be zero, but its being nonzero isn't evidence enough to assume impending failure. If the number increases, that's likely evidence the disk is dirty, scratched, or contaminated, and I'd replace it then. --
Finlay McWalterჷTalk15:27, 25 August 2012 (UTC)reply
Here's the meaning of those four numbers at the bottom:
"Normalized": A manufacturer-defined measure of healthiness where 100 (sometimes 200) is normal and smaller values are worse.
"Worst": The lowest normalized value this parameter has ever had on this drive.
"Threshold": The normalized value at which the drive should probably be replaced (according to the manufacturer).
"Value": More detailed information; in this case, the actual count of reallocated sectors.
There was a Google study a while back showing that SMART-reported errors of any sort were highly correlated with impending drive failure. If it were my drive I'd take it out of service immediately, rather than backing it up and waiting for it to fail.
69.228.170.132 (
talk)
02:25, 26 August 2012 (UTC)reply
It's not an error. The health parameter is at 100%. I don't know why this particular SMART monitor treats any nonzero raw value for this parameter as worthy of a message at all. Most wouldn't. There's a green circle next to the message, which I suppose is meant to mean that it's purely informational. --
BenRG (
talk)
04:52, 26 August 2012 (UTC)reply
Thanks everyone! I experienced a similar situation with the same computer a year ago. It then had three disks, which all were new. When I returned from summer holidays, all three disks had SMART-reported error messages similar to the one showed here. I then replaced all three disks. Given the time of year that these problems have occurred, I wonder if it may have been caused by transients on the power grid during thunderstorms. All the circles were green then, too. The message next to the circle when absolutely everything is ok, is "Disk is healthy". I also noticed now, that there is a mouseover message associated with the Reallocated Sector Count warning, that says:
Type: Failure is a sign of imminent failure (Pre-Fail).
Updates: Every time data is collected (Online)
Raw: 0x050000000000
I'm not really sure what to make of that information, i.e. if it is saying that there actually is a failure condition now, or if it is saying what the interpretation would be if there were a failure condition. To be on the safe side, my immediate action will be to switch the roles of the two 1.5 TB disks on the system (the problem disk holds the /home partition, the other 1.5 TB disk mirrors /home, rsynced every night). --
NorwegianBluetalk10:32, 26 August 2012 (UTC)reply
It's saying what the interpretation would be if there were a failure condition (i.e., if the normalized value dropped below 36). This interpretation is also chosen by the drive manufacturer. There's some information
here.
I'm kind of depressed that Google's data is convincing people to send perfectly good drives to the dump out of fear of random failure (most of their drives with a few reallocation events didn't fail). But I guess it's cheaper than RAID mirroring if downtime is expensive but not too expensive. --
BenRG (
talk)
20:04, 26 August 2012 (UTC)reply
From the Google study "Failure Trends in a Large Disk Drive Population": "As with scan errors, the presence of reallocations seems to have a consistent impact on AFR for all age groups (Figure 7), even if slightly less pronounced. Drives with one or more reallocations do fail more often than those with none. The average impact on AFR appears to be between a factor of 3-6x." AFR is Annualised Failure Rate. So it does appear that this is indicating a higher than normal liklihood of disk failure. --
Phil Holmes (
talk)
10:00, 26 August 2012 (UTC)reply
There are many standard methods to communicate to an embedded device. You are undoubtedly familiar with the USB protocol; many modern devices have hardware support for USB, so you can program and debug using USB. Another common standard is
JTAG, which specifies certain low-level communication methods and defines a hardware interface. JTAG is supported on many embedded devices, even when USB is not. Other embedded computers use less common protocols; or they use customized technology created by the manufacturer or designer.
To run a program on an embedded device, you can either boot directly to the program; or you can run a (simple)
operating system on the device. To boot directly, your program must know intricate details of the hardware, and manage the low-level device bring-up. For this reason, many embedded systems come with a ready-to-use operating system. The operating system will specify how to load and run external programs. Many new systems use full-fledged file systems that are compatible with desktop computers. Often, an embedded computer can attach an
SD card and treat it as a "hard disk" with a standard file system; the embedded operating system loads the program file and executes its main function. Other embedded devices can be programmed over JTAG, and the program may reside in RAM only until the next reboot; or can be stored to a nonvolatile memory.
Nimur (
talk)
17:39, 26 August 2012 (UTC)reply
Legality of redistributing software available for free download
What is the legality of redistributing software that's available for free download, without permission from the copyright holder? Specifically, several companies make available software programs as a large (>2 GB) DVD images for free download. I would like to republish the ISOs on my web site as BitTorrent links, to help those with unstable Internet connections or unstable computers. I understand that copyright prohibits copying and redistribution, but does it make a difference that the copyright holder is not trying to copy-restrict the ISOs themselves, instead requiring an activation key to use the software?
63.152.89.215 (
talk)
21:01, 25 August 2012 (UTC)reply
Check the license distributed with the software. It will tell you whether redistribution is allowed. Just because something is provided for no charge (gratis) does not mean it is free for you to redistribute — it does not mean that copyright does not apply, it does not mean that your own redistribution of it would not be a copyright violation. --
Mr.98 (
talk)
23:08, 25 August 2012 (UTC)reply