Ubuntu/shred?
-
@jimmy9008 said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
@jimmy9008 said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
If drives are identical in all servers, why don't you just randomly mix them? Pull drive 2 from server 1 and swap it with drive 4 from server 2, etc. Then just destroy the arrays, create new, preferably different RAID levels and just write some sample data.
Wouldn't this leave quite a risk of the data being on a drive still?
I have mixed the drives. Destroyed the arrays, and set as Raid0. Then, running shred on those new Raid 0 arrays...Risk? Unlikely. If you mix few drives from each array in few servers, there's no way to recover it unless you get the original set of drives together. The more drives and servers, the lower the chances of re-assembling the array. You're not donating these to NSA, are you?
No, lol. Two servers are going to a School to be their production environment. Another server is going to a different School to be a lab machine so students can try virtualisation.
Then just make sure complete set of disks from any server doesn't end up in one school. Schools don't have the budgets/personnel/skills/time/motivation to play the NSA.
Let me illustrate what will happen when you mix disks. In a set of 6 disks in 3 servers you have some data, but that data is completely unknown to bad actor. So:
ABCDEF - in server 1, abcdef in server 2, and 123456 in server 3. After mixing you end up with Ae2DE4 in server 1, a3BF16 in server 2 and bcCd5f in server 3. After writing some random data you'll have Ae2DEx, a3BF1y, and bcCd5z. Now go ahead and try to recover original data, not knowing what it was in first place. And do it on school's time and budget.
-
@marcinozga said in Ubuntu/shred?:
Then just make sure complete set of disks from any server doesn't end up in one school. Schools don't have the budgets/personnel/skills/time/motivation to play the NSA.
But kids in a lab do.
-
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
Then just make sure complete set of disks from any server doesn't end up in one school. Schools don't have the budgets/personnel/skills/time/motivation to play the NSA.
But kids in a lab do.
And since when kids in labs are allowed to sit there for hours swapping disks between servers?
-
@dashrender and if they are like the kids in my labs then they will be absolutely malicious just for the fun of it
-
@marcinozga said in Ubuntu/shred?:
ABCDEF - in server 1, abcdef in server 2, and 123456 in server 3. After mixing you end up with Ae2DE4 in server 1, a3BF16 in server 2 and bcCd5f in server 3. After writing some random data you'll have Ae2DEx, a3BF1y, and bcCd5z. Now go ahead and try to recover original data, not knowing what it was in first place. And do it on school's time and budget.
Is is of course true, but doesn't paint the picture accurately. Each of your above listed slots of data, represented by A and a and 1, etc are blocks of data, possibly more than 4KB worth A tone of valuable data could be stored in 4KB.
-
Eurgh... even the 0's on the array are taking ages. Been running it since lunch, 40 GiB/3.7TiB finished, 1%
-
@jmoore said in Ubuntu/shred?:
@dashrender and if they are like the kids in my labs then they will be absolutely malicious just for the fun of it
^^This. Let the kids play with it, they'll do the job for you.
-
@marcinozga said in Ubuntu/shred?:
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
Then just make sure complete set of disks from any server doesn't end up in one school. Schools don't have the budgets/personnel/skills/time/motivation to play the NSA.
But kids in a lab do.
And since when kids in labs are allowed to sit there for hours swapping disks between servers?
That's not the point, or even required. They could pull images off the drives, then mess with the data like putting paper shreddings back together if they wanted to.
My point is that when there is a will, there is a way. And you hear stories all the time about how some kid some some completely unexpected thing in school - so I would expect no less to be possible here.
-
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
Then just make sure complete set of disks from any server doesn't end up in one school. Schools don't have the budgets/personnel/skills/time/motivation to play the NSA.
But kids in a lab do.
And since when kids in labs are allowed to sit there for hours swapping disks between servers?
That's not the point, or even required. They could pull images off the drives, then mess with the data like putting paper shreddings back together if they wanted to.
My point is that when there is a will, there is a way. And you hear stories all the time about how some kid some some completely unexpected thing in school - so I would expect no less to be possible here.
Yep, agree. But by doing what I've done, its pretty safe. So, I am happy with that.
-
@marcinozga said in Ubuntu/shred?:
@jmoore said in Ubuntu/shred?:
@dashrender and if they are like the kids in my labs then they will be absolutely malicious just for the fun of it
^^This. Let the kids play with it, they'll do the job for you.
No, you're missing my point - which is that the kids might go out of their way to reconstruct the data, so they won't be doing the job of destroying the data, they will be be doing the opposite and rebuilding it - why - because they can.
What are the chances of this? Low I would say, but not anywhere near zero. And kids in a lab, it's higher than the chances that someone would try to get data off an old copy machine HD I would think.
-
@jimmy9008 said in Ubuntu/shred?:
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
Then just make sure complete set of disks from any server doesn't end up in one school. Schools don't have the budgets/personnel/skills/time/motivation to play the NSA.
But kids in a lab do.
And since when kids in labs are allowed to sit there for hours swapping disks between servers?
That's not the point, or even required. They could pull images off the drives, then mess with the data like putting paper shreddings back together if they wanted to.
My point is that when there is a will, there is a way. And you hear stories all the time about how some kid some some completely unexpected thing in school - so I would expect no less to be possible here.
Yep, agree. But by doing what I've done, its pretty safe. So, I am happy with that.
Cool - as for your format commands - why did you put the drives back into an array? Leave them all separate, and write zeros to each one independently. Also, zeros along won't protect your data. True random data is the only way to really get there, and even then, only with multiple passes. But a single track of zeros gives an attacker knowledge of what they are trying to look past, i.e. your track of zeros. By using random data, the attacker has more work to find what the previous magnetic field was.
-
@dashrender said in Ubuntu/shred?:
@jimmy9008 said in Ubuntu/shred?:
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
@dashrender said in Ubuntu/shred?:
@marcinozga said in Ubuntu/shred?:
Then just make sure complete set of disks from any server doesn't end up in one school. Schools don't have the budgets/personnel/skills/time/motivation to play the NSA.
But kids in a lab do.
And since when kids in labs are allowed to sit there for hours swapping disks between servers?
That's not the point, or even required. They could pull images off the drives, then mess with the data like putting paper shreddings back together if they wanted to.
My point is that when there is a will, there is a way. And you hear stories all the time about how some kid some some completely unexpected thing in school - so I would expect no less to be possible here.
Yep, agree. But by doing what I've done, its pretty safe. So, I am happy with that.
Cool - as for your format commands - why did you put the drives back into an array? Leave them all separate, and write zeros to each one independently. Also, zeros along won't protect your data. True random data is the only way to really get there, and even then, only with multiple passes. But a single track of zeros gives an attacker knowledge of what they are trying to look past, i.e. your track of zeros. By using random data, the attacker has more work to find what the previous magnetic field was.
Added into one array as that just made sense at the time. From the array utility, destroy the array. Then create new array raid 0 of all disks. (Just made sense). Lol.
I thought that random was great, but you are pretty much unlikely to pull anything off of the drives once zeroed... (especially if the disks were moved too), which I can still do.
-
I'm pretty late here... zeros is not faster because the speed is determined by the drive, not the system. The system can create content or just put zeroes all faster than the drive can write them to disk.
-
@scottalanmiller said in Ubuntu/shred?:
I'm pretty late here... zeros is not faster because the speed is determined by the drive, not the system. The system can create content or just put zeroes all faster than the drive can write them to disk.
I read that as the CPU doesn't have to 'think' of the random data, its faster, as its only doing 0's. That processing to generate the random data is skipped - though, like I said... I don't really know about shred etc hence asking
-
@jimmy9008 said in Ubuntu/shred?:
@scottalanmiller said in Ubuntu/shred?:
I'm pretty late here... zeros is not faster because the speed is determined by the drive, not the system. The system can create content or just put zeroes all faster than the drive can write them to disk.
I read that as the CPU doesn't have to 'think' of the random data, its faster, as its only doing 0's. That processing to generate the random data is skipped - though, like I said... I don't really know about shred etc hence asking
Right, and my point was that the CPU was not the bottleneck, so that the CPU has to do "less" doesn't change the speed of the process. Someone is assuming that this is 1980 and that disks are faster than CPUs
-
@scottalanmiller said in Ubuntu/shred?:
@jimmy9008 said in Ubuntu/shred?:
@scottalanmiller said in Ubuntu/shred?:
I'm pretty late here... zeros is not faster because the speed is determined by the drive, not the system. The system can create content or just put zeroes all faster than the drive can write them to disk.
I read that as the CPU doesn't have to 'think' of the random data, its faster, as its only doing 0's. That processing to generate the random data is skipped - though, like I said... I don't really know about shred etc hence asking
Right, and my point was that the CPU was not the bottleneck, so that the CPU has to do "less" doesn't change the speed of the process. Someone is assuming that this is 1980 and that disks are faster than CPUs
Ok, makes sense. No dates on the stuff I was reading online.
So, either way, would a zero of the entire array, then moving the disks between servers after already using array utility to destroy the array and make a raid 0... would that be pretty safe?
-
@jimmy9008 said in Ubuntu/shred?:
@scottalanmiller said in Ubuntu/shred?:
@jimmy9008 said in Ubuntu/shred?:
@scottalanmiller said in Ubuntu/shred?:
I'm pretty late here... zeros is not faster because the speed is determined by the drive, not the system. The system can create content or just put zeroes all faster than the drive can write them to disk.
I read that as the CPU doesn't have to 'think' of the random data, its faster, as its only doing 0's. That processing to generate the random data is skipped - though, like I said... I don't really know about shred etc hence asking
Right, and my point was that the CPU was not the bottleneck, so that the CPU has to do "less" doesn't change the speed of the process. Someone is assuming that this is 1980 and that disks are faster than CPUs
Ok, makes sense. No dates on the stuff I was reading online.
So, either way, would a zero of the entire array, then moving the disks between servers after already using array utility to destroy the array and make a raid 0... would that be pretty safe?
Decently safe, short of people going to forensic lengths to get a little data from the machine, there is nothing there.
-
@scottalanmiller said in Ubuntu/shred?:
@jimmy9008 said in Ubuntu/shred?:
@scottalanmiller said in Ubuntu/shred?:
@jimmy9008 said in Ubuntu/shred?:
@scottalanmiller said in Ubuntu/shred?:
I'm pretty late here... zeros is not faster because the speed is determined by the drive, not the system. The system can create content or just put zeroes all faster than the drive can write them to disk.
I read that as the CPU doesn't have to 'think' of the random data, its faster, as its only doing 0's. That processing to generate the random data is skipped - though, like I said... I don't really know about shred etc hence asking
Right, and my point was that the CPU was not the bottleneck, so that the CPU has to do "less" doesn't change the speed of the process. Someone is assuming that this is 1980 and that disks are faster than CPUs
Ok, makes sense. No dates on the stuff I was reading online.
So, either way, would a zero of the entire array, then moving the disks between servers after already using array utility to destroy the array and make a raid 0... would that be pretty safe?
Decently safe, short of people going to forensic lengths to get a little data from the machine, there is nothing there.
Ta
-
Any really critical data on those desktops? As long as it isn't like government secrets or something, I'd think that zeroing out is more than enough.
-
shredsounds better than what I normally do:dd if=/dev/random of=/dev/sda bs=4k/dev/urandom is a better RNG, but slower. Still, sounds like shred automates the process for you.
