View Full Version : logical volumes that span more than 1 disk...
really suck when one of those disks fails.
and don't any of you even think of mentioning backups. unfortunatley, I work in an environment where it just isn't practical to backup hundreds of gigabytes of data per person on a regular basis. In other words, there's nowhere to put backups of anything except the small amout of stuff you are allowed to put in your home directory. :mad: :mad: :mad: :mad: :mad:
I wasn't thinking backup. I was thinking raid.
um, yeah that's bad
disks are pretty cheap these days... comparatively
RAID 5 or 1+0 is much better
Stoop Dawg
4/4/2006, 06:07 PM
really suck when one of those disks fails.
and don't any of you even think of mentioning backups. unfortunatley, I work in an environment where it just isn't practical to backup hundreds of gigabytes of data per person on a regular basis. In other words, there's nowhere to put backups of anything except the small amout of stuff you are allowed to put in your home directory. :mad: :mad: :mad: :mad: :mad:
Can't you just fire off one of those little particles really fast and have it go back in time and get it for you?
Sheesh, you scientists.....
Edit: And, uh, yeah, RAID is where you wanna be. Hot-swappable, preferrably.
Vaevictis
4/4/2006, 06:49 PM
This is why most people don't span multiple disks without mirroring or parity of some kind. Doesn't help you now, I know, but still... stripes across mirrors > *.
slickdawg
4/4/2006, 07:44 PM
Well, you could do what I did, and build a 60 terabyte raid array
that features multiple 2GB fiber channel trays of disks configured
in raid 5 and raid 1+0. Then you have that backed up on an HSM
that is 2.2 petabytes in size and scale up to about 100PB with
no problems.
just sayin'
:D
we have some raid arrays, and they have partially saved my *** because the bulk of what I have done for the past 3 years on there, but since we get our money from the gub-ment, we can't make every machine a raid array. limited funds + large need for disk space mean that we can't make as many raid arrays as we would really like to. the fact of the matter is that this is just simply the environment we have to work in.
NormanPride
4/4/2006, 08:03 PM
Well, you could do what I did, and build a 60 terabyte raid array
that features multiple 2GB fiber channel trays of disks configured
in raid 5 and raid 1+0. Then you have that backed up on an HSM
that is 2.2 petabytes in size and scale up to about 100PB with
no problems.
just sayin'
:D
Man, you could fit SO MUCH PR0N ON THERE. :texan:
Vaevictis
4/4/2006, 08:29 PM
we have some raid arrays, and they have partially saved my *** because the bulk of what I have done for the past 3 years on there, but since we get our money from the gub-ment, we can't make every machine a raid array.
Me, whenever I would set up machines, I'd calculate how much it would cost to make them appropriately redundant and give send an email to the boss with the price tag. I'd also give him the stripped down version, and an estimate of how long it would take to bring the machine up without the data, and remind him that failed disks go offsite for weeks to specialists who charge thousands of dollars to recover a random amount of data.
At that point, it's a business decision, and if the funds aren't made available to assure continuity, it ain't my f*cking problem.
Out of curiousity, have you considered centralized storage or networked storage? HA NAS/SAN stuff is really getting quite affordable, and if you can't afford that, you can always ghetto rig networked solutions like OpenAFS on spare disk space across your network.
Vaevictis
4/4/2006, 08:35 PM
Another option: Pick someone at work you don't like. Automate their job. Use the funds released by their being laid off to buy the hardware you need.
(This is always a happy option if you can pull it off; not only does it get rid of someone you don't like, but you get to buy neat toys that make YOUR job easier. And it makes you look good to the boss, too.)
slickdawg
4/4/2006, 08:37 PM
Man, you could fit SO MUCH PR0N ON THERE. :texan:
In a 1st generation array that held 7TB, about the time Napster was at its
peak, I used to tell folks "everything on Napster would fit on here right now"
That just freaked folks out - they could comprehend that to a degree.
slickdawg
4/4/2006, 08:38 PM
Me, whenever I would set up machines, I'd calculate how much it would cost to make them appropriately redundant and give send an email to the boss with the price tag. I'd also give him the stripped down version, and an estimate of how long it would take to bring the machine up without the data, and remind him that failed disks go offsite for weeks to specialists who charge thousands of dollars to recover a random amount of data.
At that point, it's a business decision, and if the funds aren't made available to assure continuity, it ain't my f*cking problem.
Out of curiousity, have you considered centralized storage or networked storage? HA NAS/SAN stuff is really getting quite affordable, and if you can't afford that, you can always ghetto rig networked solutions like OpenAFS on spare disk space across your network.
Man, I don't know about trusting AFS. My experience in the 90's was BAD.
Vaevictis
4/4/2006, 08:42 PM
Man, I don't know about trusting AFS. My experience in the 90's was BAD.
I don't really disagree. But if your option is trying to use AFS (which you can now get for free) to mirror data in case of failure or not trying at all, I'm thinking that if you can spare the time, AFS is the better choice.
slickdawg
4/4/2006, 08:45 PM
The one thing AFS has going for it is that it's free and available for many platforms.
solstice disk suite works well, but it's Solaris only
Veritas works great, but you pay out the a$$
Me, whenever I would set up machines, I'd calculate how much it would cost to make them appropriately redundant and give send an email to the boss with the price tag. I'd also give him the stripped down version, and an estimate of how long it would take to bring the machine up without the data, and remind him that failed disks go offsite for weeks to specialists who charge thousands of dollars to recover a random amount of data.
At that point, it's a business decision, and if the funds aren't made available to assure continuity, it ain't my f*cking problem.
Out of curiousity, have you considered centralized storage or networked storage? HA NAS/SAN stuff is really getting quite affordable, and if you can't afford that, you can always ghetto rig networked solutions like OpenAFS on spare disk space across your network.
NFS across the network, most storage is distributed amongst desktops.
this is a very different envioronment that we work in. the feds kick in only so much money for the lab, and only so much of that goes towards computing. they mostly own the raid arrays, but those aren't directly used for analysis.
but the bulk of the computers at the lab (and the ones used for analysis) are actually owned by the individual universities at the lab. so OU owns 7. well, now 5, of the computers at the lab, and 4 of those have 2 120 gig hard disks where we store our temporary work. the other one is a raid5 array, but only has so much space. so we can't store very many files that contain billions of picked events there. those mostly go in our /scratch areas.
on the raid array we try to put the code that generates those files of events, and thats pretty much what I have there. Although some of my code was on the local machine that died, but there are old versions of it in other places.
but it takes several weeks to re-generate many of the files that were lost due to the disk failure.
Funding for this field for the last several years has declined ever since the SSC died. the crunch is mostly felt at the university level. meaning that the OU HEP group doesn't have a lot of money to invest in making every computer mirrored. It's impractical when the thing we most have a need for is space.
and this is my frickin problem, primarily because my thesis depends on a lot of what was on there. Yes, I can re-create most of that, but I really don't want to.
so the taxpayers have the choice. you can pay for more redundancy, or you can pay mine, and many other graduate students, salaries for an extra month or so. ;)
sooner_born_1960
4/4/2006, 08:57 PM
I try to always have a couple floppies in my shirt pocket in case I need to backup my data.
Vaevictis
4/4/2006, 09:06 PM
NFS across the network, most storage is distributed amongst desktops.
Are those desktops like most I've encountered, in that there tends to be a lot of unused space on them individually (so long as you b*tchslap the lusers into behaving properly)?
If so, there's got to be a way to make use of that unused space, it's just a matter of finding the time to make it happen.
Are those desktops like most I've encountered, in that there tends to be a lot of unused space on them individually (so long as you b*tchslap the lusers into behaving properly)?
If so, there's got to be a way to make use of that unused space, it's just a matter of finding the time to make it happen.
some are some aren't. depending on what you are doing, you sometimes have to search for disk space. Its not uncommon to temporarily borrow a disk that belongs to another university until you can make some space on your own machine.
and the /home directories (the only ones that get backed up) are not distributed. those reside in some central storage location, but they also have pretty rigid quotas.
but now we've gotten way far away from the original intent of this thread, which is that LVM sucks ***. I have no idea why anybody decided that we should use it.
Sooner_Bob
4/4/2006, 09:59 PM
I had a 2 disk array setup that crashed on me a while back. Luckily, I was able to use File Scavenger to rebuild and recover my data.
I couldn't even imagine a terabyte or petabyte drive. That would rock.
SicEmBaylor
4/4/2006, 10:33 PM
Could you just use a floppy disk?
Vaevictis
4/4/2006, 10:42 PM
but now we've gotten way far away from the original intent of this thread, which is that LVM sucks ***. I have no idea why anybody decided that we should use it.
LVM is awesome, if you use it right. Spanning disks without redundancy below it is not using it right.
The ability to take snapshots is awesome. The ability to use LVM to span arrays is awesome. The ability to transparently move volumes from one disk to another is awesome. In short, the fact that your data volumes are no longer directly tied to your hard disks is *awesome*... if you don't do something silly like span across non-redundant physical devices and expect it *not* to bite you in the ***.
slickdawg
4/4/2006, 10:42 PM
I had a 2 disk array setup that crashed on me a while back. Luckily, I was able to use File Scavenger to rebuild and recover my data.
I couldn't even imagine a terabyte or petabyte drive. That would rock.
It's extremely cool, but unfortunately, if something does go wrong, it's
a huge ****ing mess.
LVM is awesome, if you use it right. Spanning disks without redundancy below it is not using it right.
The ability to take snapshots is awesome. The ability to use LVM to span arrays is awesome. The ability to transparently move volumes from one disk to another is awesome. In short, the fact that your data volumes are no longer directly tied to your hard disks is *awesome*... if you don't do something silly like span across non-redundant physical devices and expect it *not* to bite you in the ***.
and these are all things that Im sure could be nice. but I just want my computers to work, and not to have to waste my time becomming the IT department because somebody else set things up this way three years ago. The fact that the non-workage of one disk saps the other, makes the choice of even using LVM in the first place, stupid. I don't need that ****. I need a machine that compiles and runs my damn code.
sooner_born_1960
4/4/2006, 11:24 PM
Could you just use a floppy disk?
That's just dumb. ;)
Vaevictis
4/4/2006, 11:43 PM
but I just want my computers to work, and not to have to waste my time becomming the IT department because somebody else set things up this way three years ago. The fact that the non-workage of one disk saps the other, makes the choice of even using LVM in the first place, stupid. I don't need that ****. I need a machine that compiles and runs my damn code.
I understand completely that you're frustrated, but the problem is not with LVM. Used properly, it is a powerful tool that can make your life infinitely easier. Used improperly, it is a powerful tool with which to blow your own foot off.
In this case, the fault lies not with LVM, but rather with whoever set it up. It did *exactly* what it was supposed to do; it spanned the volume across multiple disks. Anyone who knows about system administration would know that this increases the amount of data lost in the event of disk failure. It's basically the first thing you learn about when you learn about disk arrays. If the risk of this kind of failure outweighed the benefits received by doing it this way, then the installer should *not* have configured it that way.
This whole, "I just want it to work" thing is, imho, a bogus (but understandable) attitude. These computers are obviously tools of your trade. It would be nice if the tools just did the work for you. But they don't. If you don't know what's going on with the tools of your trade, you are just begging for trouble.
I understand completely that you're frustrated, but the problem is not with LVM. Used properly, it is a powerful tool that can make your life infinitely easier. Used improperly, it is a powerful tool with which to blow your own foot off.
In this case, the fault lies not with LVM, but rather with whoever set it up. It did *exactly* what it was supposed to do; it spanned the volume across multiple disks. Anyone who knows about system administration would know that this increases the amount of data lost in the event of disk failure. It's basically the first thing you learn about when you learn about disk arrays. If the risk of this kind of failure outweighed the benefits received by doing it this way, then the installer should *not* have configured it that way.
This whole, "I just want it to work" thing is, imho, a bogus (but understandable) attitude. These computers are obviously tools of your trade. It would be nice if the tools just did the work for you. But they don't. If you don't know what's going on with the tools of your trade, you are just begging for trouble.
the problem may not be *with* LVM, but what I've been trying to say is that it is with the fact that LVM was used in the first place. It was completely unnessecary.
Im only as much of an IT guy as I have to be. theres way too much crap to keep up with in that department that doing so would keep me from being able to do my job. The fact that I now have to learn as much about LVM as I can just to try to save what I can from the working disk on that computer is more than I really need to know. The one person in our collaboration who has done this before has left. So, it falls to me.
Vaevictis
4/5/2006, 12:00 AM
the problem may not be *with* LVM, but what I've been trying to say is that it is with the fact that LVM was used in the first place. It was completely unnessecary.
Ah hah. I misunderstood. (I think my misunderstanding that is fair considering your statement that "LVM sucks ***" ;) ) Well, that said, you do realize that some kind of LVM is used by default by most modern Unixes, right? Its use is not really so much of a choice anymore.
The fact that I now have to learn as much about LVM as I can just to try to save what I can from the working disk on that computer is more than I really need to know. The one person in our collaboration who has done this before has left. So, it falls to me.
It just occured to me that it *may* be possible to just resize the volume and overlaid filesystem to the size of one disk, *if* the volume was a simple linear concatanation. The LVM volume should have no trouble, in any case. Depending on the filesystem used, it may be able to handle that behavior gracefully. Or not. OTOH, if it was a stripe set, give up now. From what you said, you'll be able to regenerate the data before you'll ever be able to reconstruct anything useful.
Ah hah. I misunderstood. (I think my misunderstanding that is fair considering your statement that "LVM sucks ***" ;) ) Well, that said, you do realize that some kind of LVM is used by default by most modern Unixes, right? Its use is not really so much of a choice anymore.
It just occured to me that it *may* be possible to just resize the volume and overlaid filesystem to the size of one disk, *if* the volume was a simple linear concatanation. The LVM volume should have no trouble, in any case. Depending on the filesystem used, it may be able to handle that behavior gracefully. Or not. OTOH, if it was a stripe set, give up now. From what you said, you'll be able to regenerate the data before you'll ever be able to reconstruct anything useful.
bingo. as far as I know, there is no striping going on...thank god. but, it turns out apparently, that the resizing only works if LVM2 was used. since I didn't set up the machine, I have no idea if this is using LVM2 or LVM1. In fact, I have no idea (yet) how to even tell. further, its possible that the disk that failed might be sent to a data recovery center (because someone higher up the totem pole than I had a lot of stuff there that really mattered to them a lot). so if I resize the working disk in order to read it, I don't yet know if Im going to screw up potential data recovery efforts, which I think will succeed, since Im 90% sure that the damage to the disk was contained to the electronics. Right now the only question is the cost. And of course, how much damage I can do in the meantime.
and these are all things that Im sure could be nice. but I just want my computers to work, and not to have to waste my time becomming the IT department because somebody else set things up this way three years ago. The fact that the non-workage of one disk saps the other, makes the choice of even using LVM in the first place, stupid. I don't need that ****. I need a machine that compiles and runs my damn code.
if it was me and my time i was talking about i'd be buying a couple 250GB drives and mirroring them. they are like $100 a piece
i would think that whoever has to pay you for an extra month now could find $200
this doesn't help you get the old data though :D
if it was me and my time i was talking about i'd be buying a couple 250GB drives and mirroring them. they are like $100 a piece
i would think that whoever has to pay you for an extra month now could find $200
this doesn't help you get the old data though :D
that would be you, dear taxpayer. fork it over! :D
Vaevictis
4/5/2006, 12:29 AM
but, it turns out apparently, that the resizing only works if LVM2 was used. since I didn't set up the machine, I have no idea if this is using LVM2 or LVM1. In fact, I have no idea (yet) how to even tell.
Okay, then. That sounds like Linux, yes? I'm guessing RedHat (or fedora), since it's one of the more popular ones. If so:
rpm -qa | grep -i lvm
You will probably see something that looks like this:
system-config-lvm-0.9.32-1.0
lvm2-2.01.08-2.1
Pick the one that looks like the lvm package, and run an RPM query on it. In my case:
rpm -qi lvm2
Read the description. You should know if what you're looking at is the right thing at this point.
If not, try:
rpm -ql lvm2
This will give a list of files associated with the package. Chances are, there will be some documentation; you now know what files to poke at and probably what man pages to read.
further, its possible that the disk that failed might be sent to a data recovery center (because someone higher up the totem pole than I had a lot of stuff there that really mattered to them a lot). so if I resize the working disk in order to read it, I don't yet know if Im going to screw up potential data recovery efforts, which I think will succeed, since Im 90% sure that the damage to the disk was contained to the electronics.
WARNING WARNING WARNING: Before you do any of the below, double check and verify for yourself. I think it'll work, but this is *your* *** on the line, so cover it yourself.
If you have a spare disk of the identical size and dimensions, it should be possible to make a raw copy of it:
dd if=/dev/hda of=/dev/hdb bs=1M
Obviously, adjust the device files appropriately, and DONT F*CK UP by reversing the order and copying *to* the drive you're trying to copy *from*.
If you don't have an identical disk, this may also work with a disk of larger size, but I don't know that for a fact; if it doesn't, if you do have a larger one, it should *also* be possible to copy to a file and register it as a loopback device, then mount it up that way -- a process which I have not done in awhile, and so I'll have to leave to you to figure out.
that would be you, dear taxpayer. fork it over! :D
heh
i gave at the office
SicEmBaylor
4/5/2006, 12:33 AM
That's just dumb. ;)
I know ;-)
Vaevictis
4/5/2006, 12:37 AM
i would think that whoever has to pay you for an extra month now could find $200
This is why I used to do the thing I described above regarding the business decision. This is the point where I rub salt into the wound so that the business type NEVER forgets the 100 pounds of cure he had to spend when he could have paid for an ounce of prevention months or years before.
It's not a matter of if hard drives fail. It's a matter of when. If you use a hard drive, hard drive failure is not a possibility -- it is a *certainty.* The only question is how much it's going to hurt when it happens.
Okay, then. That sounds like Linux, yes? I'm guessing RedHat (or fedora), since it's one of the more popular ones. If so:
yeah, fedora...core 2 I think, but I'm not sure about that one.
rpm -qa | grep -i lvm
You will probably see something that looks like this:
system-config-lvm-0.9.32-1.0
lvm2-2.01.08-2.1
Pick the one that looks like the lvm package, and run an RPM query on it. In my case:
rpm -qi lvm2
Read the description. You should know if what you're looking at is the right thing at this point.
If not, try:
rpm -ql lvm2
This will give a list of files associated with the package. Chances are, there will be some documentation; you now know what files to poke at and probably what man pages to read.
unfortunately, I can't do this on the machine that experienced the disk failure. the master disk was the one that failed, and if its hooked up, neither disk is detected. the slave disk has no mount point because it was entirely one part of a logical volume that spanned disks, so I can't mount that disk.
However, if I do this on that machines identical twin (purchased and set up at the same time), I see that this is in fact LVM1, which I am told does not support re-sizing of logical volumes. crappy McCrapperPants. Somebody should really die because of this.
WARNING WARNING WARNING: Before you do any of the below, double check and verify for yourself. I think it'll work, but this is *your* *** on the line, so cover it yourself.
If you have a spare disk of the identical size and dimensions, it should be possible to make a raw copy of it:
dd if=/dev/hda of=/dev/hdb bs=1M
Obviously, adjust the device files appropriately, and DONT F*CK UP by reversing the order and copying *to* the drive you're trying to copy *from*.
If you don't have an identical disk, this may also work with a disk of larger size, but I don't know that for a fact; if it doesn't, if you do have a larger one, it should *also* be possible to copy to a file and register it as a loopback device, then mount it up that way -- a process which I have not done in awhile, and so I'll have to leave to you to figure out.
I may just try that. One of our machines was sacrificed this week so that its power supply could be used to save another machine that people are actually using (the one that was sacrificed crashed a lot and was really not used by anyone). Meaning that there are 2 120 GB drives sitting there not doing much of anything....
Vaevictis
4/5/2006, 01:08 AM
However, if I do this on that machines identical twin (purchased and set up at the same time), I see that this is in fact LVM1, which I am told does not support re-sizing of logical volumes. crappy McCrapperPants. Somebody should really die because of this.
Boo. (And yeah, silly me for giving commands to run on a dead computer.)
Hmm. Since you have an identical twin, here's an idea that might lead somewhere. If you have an identical hard drive, you *may* be able to copy the LVM metadata off of the identical twin (once you figure out where it is on the physical disk -- it should be the same -- you can use the "dd" command for this too, read the man page) and onto a new hard drive, slap it in, start up the LVM subsystem and resize the ext...3 (?) filesystem.
Because the LVM doesn't know anything about the filesystem, it won't care that the filesystem isn't actually there; all it cares about is its own stuff.
Boo. (And yeah, silly me for giving commands to run on a dead computer.)
Hmm. Since you have an identical twin, here's an idea that might lead somewhere. If you have an identical hard drive, you *may* be able to copy the LVM metadata off of the identical twin (once you figure out where it is on the physical disk -- it should be the same -- you can use the "dd" command for this too, read the man page) and onto a new hard drive, slap it in, start up the LVM subsystem and resize the ext...3 (?) filesystem.
Because the LVM doesn't know anything about the filesystem, it won't care that the filesystem isn't actually there; all it cares about is its own stuff.
ooooh...thats an idea. one that will **** off the postdoc that works at that computer. But he's an *** anyways. I like it already. and yeah, it is ext3
Vaevictis
4/5/2006, 08:02 AM
Hmm. Since you have an identical twin, here's an idea that might lead somewhere. If you have an identical hard drive, you *may* be able to copy the LVM metadata off of the identical twin
After sleep, it occured to me that this may not work due to the LVM volume number. I don't know for sure. For this to work, you may have to figure out how to edit that meta data and munge it to have the same volume ID. (If this is necessary, I'm not certain it will work -- I don't really know what kind of information is contained in that volume number)
Another idea is to find out where the ext3 filesystem begins on the hard drive and copy it to a file and mount that as a loopback device. This assumes that the LVM data is contained only at the start of a partition instead of scattered throughout.
Partial Qualifier
4/5/2006, 08:28 AM
I realize Ike's funds are too limited for this option, but thought it worth mentioning here: recovery service providers can retrieve data from raid & spanned volumes. It's hella expensive though.
slickdawg
4/5/2006, 08:43 AM
I know ;-)
Nah, this would be dumb:
log in as root and type
rm -rf /*
slickdawg
4/5/2006, 08:47 AM
I realize Ike's funds are too limited for this option, but thought it worth mentioning here: recovery service providers can retrieve data from raid & spanned volumes. It's hella expensive though.
Patting my own back for a minute, after Katrina flooded out and heavily
damaged the Hancock County Sheriff's office and their entire IT
infrastruature, I went through every machine for them and retrieved
about 40% of all of the data on all of the systems. Luckily, of those that I
saved the data from, they were the most critical ones, such as the database
of who is in the jail, the bookings, bail/bond, etc, their entire ID system.
Someone like ontrack.com would have $crewed them blind.
:texan:
Stoop Dawg
4/5/2006, 10:33 AM
This whole, "I just want it to work" thing is, imho, a bogus (but understandable) attitude. These computers are obviously tools of your trade. It would be nice if the tools just did the work for you. But they don't. If you don't know what's going on with the tools of your trade, you are just begging for trouble.
Wrong. Do race car drivers work on their own cars? Do truck drivers perform maintenance on their trucks? Do *you* know how to fix everything that could go wrong on your car (assuming you require a car to get to your job every day)?
The job of IT is to make sure the computers work and there is minimal loss of productivity of the people using those computers. There is, of course, an upper bound on what IT can accomplish - and that upper bound is set by the person with the $$$.
Now, if management won't spring for redundancy, then the user is welcome to be ****ed at management instead of IT. But in no case is it the *user's* fault that IT and/or management couldn't get it right. Just my opinion, of course.
Vaevictis
4/5/2006, 11:16 AM
Wrong. Do race car drivers work on their own cars? Do truck drivers perform maintenance on their trucks? Do *you* know how to fix everything that could go wrong on your car (assuming you require a car to get to your job every day)?
I don't know any race car drivers, so I can't say. I do have some family who are truck drivers, and you're goddamn right they perform maintenance on their trucks. I don't need to know how to fix everything that can go wrong with my car. When my car breaks down, I call AAA to come tow it away, and then take the bus, take a taxi, or rent a car. Cars are replacable.
But if I have something that's critical to my job that I can't afford to lose, then I'm going to do everything within my power to make sure I don't lose it -- and that includes learning how it works and how to maintain it.
The job of IT is to make sure the computers work and there is minimal loss of productivity of the people using those computers. (...)
Now, if management won't spring for redundancy, then the user is welcome to be ****ed at management instead of IT. But in no case is it the *user's* fault that IT and/or management couldn't get it right. Just my opinion, of course.
If you read what this guy is doing, it should tell you that he IS what passes for IT in his shop. "Users" don't try to recover data from failed hard drives.
I agree that it's different if there's an RP that one can trust to take care of the job, but in this case, there ain't.
I don't know any race car drivers, so I can't say. I do have some family who are truck drivers, and you're goddamn right they perform maintenance on their trucks. I don't need to know how to fix everything that can go wrong with my car. When my car breaks down, I call AAA to come tow it away, and then take the bus, take a taxi, or rent a car. Cars are replacable.
But if I have something that's critical to my job that I can't afford to lose, then I'm going to do everything within my power to make sure I don't lose it -- and that includes learning how it works and how to maintain it.
If you read what this guy is doing, it should tell you that he IS what passes for IT in his shop. "Users" don't try to recover data from failed hard drives.
I agree that it's different if there's an RP that one can trust to take care of the job, but in this case, there ain't.
sort of. I'm a user first, and IT is much much lower on my priority level. we have this group of people that are supposed to be higher than me on the IT totempole (we call them the cluster admins). They've thrown up their hands. But they too are users first. Nobody on our collaboration has a computing degree. Everyone is a physicist first.
After sleep, it occured to me that this may not work due to the LVM volume number. I don't know for sure. For this to work, you may have to figure out how to edit that meta data and munge it to have the same volume ID. (If this is necessary, I'm not certain it will work -- I don't really know what kind of information is contained in that volume number)
Another idea is to find out where the ext3 filesystem begins on the hard drive and copy it to a file and mount that as a loopback device. This assumes that the LVM data is contained only at the start of a partition instead of scattered throughout.
the problem being that the entire second disk is the last piece of the LVM that spans disks. so there is no other partition/volume on it.
but I remember reading somewhere that the thing you described is possible if you can figure out how to change the UUID (I think that was the name...things at this level are fairly new to me nowadays) of the new drive.
Vaevictis
4/5/2006, 12:19 PM
the problem being that the entire second disk is the last piece of the LVM that spans disks. so there is no other partition/volume on it.
I'm not sure I understand what you're saying here.
but I remember reading somewhere that the thing you described is possible if you can figure out how to change the UUID (I think that was the name...things at this level are fairly new to me nowadays) of the new drive.
I usually just call it the "volume ID", but yeah, I think the UUID is what it is.
I'm not sure I understand what you're saying here.
nevermind. after looking at your post again, its clear that I misunderstood. like I said before, a lot of this is very new to me.
update:
we sent the drives to seagate because the group decided that the postdoc who had most of her unbacked up data and code on it was worth it. Seagate sent the drives back to us telling us that there was nothing they could do about it. So now I am back to square 1 (ok, I'm not, I already have all my code that was lost re-created from older versions) and trying to see if I can use the original computers identical twin to recover the data on the working drive. this could be fun.
Partial Qualifier
5/8/2006, 04:06 PM
We've had pretty good luck with these people, Ike:
http://www.vogon.us/
I'm thinking the plan to use another computer to attempt data retrieval off the one good drive will be a lesson in frustration.
We've had pretty good luck with these people, Ike:
http://www.vogon.us/
I'm thinking the plan to use another computer to attempt data retrieval off the one good drive will be a lesson in frustration.
it may be...but we'll see. They don't force you to listen to poetry do they?
Powered by vBulletin® Version 4.2.1 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.