Migration from Windows10 to Debian10

New to Debian (Or Linux in general)? Ask your questions here!
Post Reply
Message
Author
e151280
Posts: 5
Joined: 2019-09-06 00:09
Has thanked: 2 times

Migration from Windows10 to Debian10

#1 Post by e151280 »

Microsoft Windows 7 & 10 has been the primary OS on this pc since I built it in early 2010, but over the last 9 months, Microsoft has been having more troubles with their Operating Systems than I care to deal with. I had been putting off pulling the trigger on migrating to Linux, but the pain threshold has been reached. Windows update (20H1) in November 2020 completely hosed my system, and Microsoft Tech support could not solve the problem, advising me to wipe, and reinstall the OS. Then AGAIN in May 2021 another big update (20H2) that took 28 hours to apply was the last straw. I began reviewing Linux distros, to make a decision on which way to go.

I’d long been a fan of CentOS, but it is going away. Second on my list was SuSE (German Engineering), and third was Debian. I played with SuSe Tumbleweed and Debian Vms for a while, Debian won.

The Plan is to remove the 500GB Windows 10 drive, create a RAID 0 array across the two 480GB SSDs for a speedy boot volume, remove the DVD/RW optical drive from the mix, add a new 1.5TB storage volume for VMs, and keep the two 3TB storage volumes with my multimedia content.

While creating a new array on the motherboard’s embedded Intel Matrix array controller is easy, and installng Debian across the two 480GB drives in RAID 0 is pretty straightforward, I do have a concern about my multimedia volumes.

Will Debian 10 mount and recognize the two 3TB NTFS volumes? I definitely don’t want to risk losing all of that content.

I have already run a P2V against the Windows boot drive, to create a VHD, so I can keep Win10 in a VM for cases where Windows is the only option, but I want Debian Linux as my base OS for stability.

Current Hardware configuration

Gigabyte GA X58 USB3 system board with Embedded Intel Chipset SATA RAID Controller embedded Realtek Audio ALC892 disabled
Intel i7-960 64bit 8 Core 3.2GHz CPU Socket FCLGA1366 operating @ 3326 MHZ
16,384 MB DDR3 ram running @1066MHZ
500GB hard disk ( WD5000AAKX-606U6AA0 ) – Windows 10 boot volume
2x 1.5TB hard disk in RAID 0 ( 3TB NTFS Data volume for Multimedia Video files )
1x 3.0TB hard disk in RAID 0 (WD30EZRX-00D8PB0) ( 3TB NTFS Data volume for Multimedia Audio files )
ASUS DRW-24B1ST DVD/RW
AMD Radeon HD 5700 Series PCIe x16
Dual Acer S230HL (DVI-D)
Realtek RTL8111E embedded PCIe GbE Family Controller
Intel PRO/1000 GT Desktop Adapter
CreativeLabs SB X-Fi PCI
Logitech M510 Wireless Mouse
Keytronic 101key PS2 Keyboard

Proposed Hardware configuration – changes/additions in bold

Gigabyte GAX58 USB3 system board with Embedded Intel Chipset SATA RAID Controller embedded Realtek Audio ALC892 disabled
Intel i7-960 64bit 8 Core 3.2Ghz CPU Socket FCLGA1366 operating @ 3326 MHZ
16,384 MB DDR3 ram running @1066MHZ

2x 480GB SSD striped in RAID 0 (PNY Optima SSD7SC480GOPT-RB ) – Debian 10 1TB boot volume
1.5TB hard disk (New storage volume for Virtual Machines)


2x 1.5TB hard disk striped in RAID 0 ( 3TB NTFS Data volume for Multimedia Video files )
1x 3.0TB hard disk in RAID 0 (WD30EZRX-00D8PB0) ( 3TB NTFS Data volume for Multimedia Audio files )
AMD Radeon HD 5700 Series PCIe x16
Dual Acer S230HL (DVI-D)
Realtek RTL8111E embedded PCIe GbE Family Controller
Intel PRO/1000 GT Desktop Adapter
CreativeLabs SB X-Fi PCI
Logitech M510 Wireless Mouse
Keytronic 101key PS2 Keyboard
Q: What's the difference between Roast Beef, and Pea Soup?
A: It's a pretty safe bet, that ANYONE can roast beef.

This electronic message has been brought to you by the language c, and the number F.

User avatar
Hallvor
Posts: 1258
Joined: 2009-04-16 18:35
Location: Kristiansand, Norway
Has thanked: 6 times
Been thanked: 16 times

Re: Migration from Windows10 to Debian10

#2 Post by Hallvor »

Will Debian 10 mount and recognize the two 3TB NTFS volumes? I definitely don’t want to risk losing all of that content.
You can mount, read and write to NTFS. Not a problem. I can't remember of they are mounted automatically, but it's a trivial task to set everything up. Just make sure you have the ntfs-3g package.

Code: Select all

# apt install ntfs-3g
Lenovo ThinkPad T440S, Intel Core i7-4600U CPU @ 2.10GHz, 8 GB RAM, 256 GB SSD, Debian Bullseye (KDE)
Lenovo ThinkPad X240, Intel Core i5-4300U CPU @ 2.90GHz, 8 GB RAM, 120 GB SSD, Debian Bullseye (KDE)

CwF
Posts: 1189
Joined: 2018-06-20 15:16
Has thanked: 2 times
Been thanked: 6 times

Re: Migration from Windows10 to Debian10

#3 Post by CwF »

This is going to be fun!

I'd "de-windowize" your thinking.
Scrap the raid-0 plans, IMHO.
That cpu is a quad core, and with 16GB vm's will be a little tight.

It's a long road to convert. The major conversion is in your expectations. They will be a moving target for some time to come...Windows people seem to over think things. KISS

User avatar
wizard10000
Posts: 237
Joined: 2019-04-16 23:15
Location: southeastern us
Has thanked: 28 times
Been thanked: 10 times

Re: Migration from Windows10 to Debian10

#4 Post by wizard10000 »

CwF wrote: 2021-07-26 14:22...Scrap the raid-0 plans, IMHO.
Agree completely; Debian boots quickly enough on an SSD on bare metal and the RAID0 brings a second point of failure with little or no added benefit.
we see things not as they are, but as we are.
-- anais nin

e151280
Posts: 5
Joined: 2019-09-06 00:09
Has thanked: 2 times

Re: Migration from Windows10 to Debian10

#5 Post by e151280 »

Dear fellow Debian forum users,

thank you for the "ntfs-3g" clue for maintaining access to my multimedia storage volume. Btw, I see 2 recommendations to NOT create a ssd raid boot volume. I was thinking that while SSD is fast, 2 SSD drives in RAID 0 would be even faster (simultaneous reads). What danger am I not seeing? I also have a battery backed Adaptec 6805 RAID card from an old server that I can add in, if the motherboard's Intel Matrix RAID controller is seen as a "fake raid".

Please enlighten me... I am ALWAYS willing to learn.
Q: What's the difference between Roast Beef, and Pea Soup?
A: It's a pretty safe bet, that ANYONE can roast beef.

This electronic message has been brought to you by the language c, and the number F.

User avatar
wizard10000
Posts: 237
Joined: 2019-04-16 23:15
Location: southeastern us
Has thanked: 28 times
Been thanked: 10 times

Re: Migration from Windows10 to Debian10

#6 Post by wizard10000 »

e151280 wrote: 2021-07-26 18:33What danger am I not seeing? I also have a battery backed Adaptec 6805 RAID card from an old server that I can add in, if the motherboard's Intel Matrix RAID controller is seen as a "fake raid".
Yup - That Intel controller is fakeraid.

RAID0 will double your seek time, add a second point of potential failure and when used with SSD will really not give you a whole lot in return.

Also, you get zero benefit from streaming multimedia on RAID0 because any modern hard drive can more than keep up with streaming requirements - but again, RAID0 will double your seek time. Also again, second point of potential failure :)

edit: There's a difference between wants and requirements; this is your hardware and completely your business but my humble opinion is that you'd just be wasting hard drives for almost zero increase in performance :)

cheers -
we see things not as they are, but as we are.
-- anais nin

User avatar
dilberts_left_nut
Global Moderator
Global Moderator
Posts: 5148
Joined: 2009-10-05 07:54
Location: enzed
Has thanked: 1 time
Been thanked: 4 times

Re: Migration from Windows10 to Debian10

#7 Post by dilberts_left_nut »

I definitely don’t want to risk losing all of that content.
You have backups right?
If not, you already don't care about losing it, especially on a RAID0.
AdrianTM wrote:There's no hacker in my grandma...

User avatar
wizard10000
Posts: 237
Joined: 2019-04-16 23:15
Location: southeastern us
Has thanked: 28 times
Been thanked: 10 times

Re: Migration from Windows10 to Debian10

#8 Post by wizard10000 »

I have a story about one of my own lessons learned.

Built a desktop PC with dual Pentium 2 chips and hardware RAID5; three 9GB 10k rpm enterprise drives and an Adaptec wide SCSI RAID controller. Spent over $1000 on the RAID setup alone and this was 20 years ago.

Needed more storage and bought a 40GB 7200 rpm Hitachi IDE drive that ran circles around my very cool but very expensive RAID array. That's when I learned that seek time matters and sector density is a much better indicator of disk throughput than rotational speed will ever be. I sold the RAID card and drives on eBay for about half what I paid for them :)

edit for accuracy as my almost-senior citizen brain remembers this machine: These were Pentium 3 chips, not Pentium 2. I did build a dual processor Pentium 2 but remember now that these P2 were slot 1 processors on a legendary Asus P2B-DS motherboard. I skipped the original Pentium as I'd overclocked a 133MHz AMD 486 processor to 160MHz and used it to embarrass my friend and his shiny new Pentium 75. Man, I'm getting old :)
we see things not as they are, but as we are.
-- anais nin

User avatar
ticojohn
Posts: 1080
Joined: 2009-08-29 18:10
Location: /home/heart
Has thanked: 5 times
Been thanked: 17 times

Re: Migration from Windows10 to Debian10

#9 Post by ticojohn »

wizard10000 wrote: 2021-07-27 12:31 Man, I'm getting old :)
Yeah, me too (72). But I think it's better than the alternative. :mrgreen:
I am not irrational, I'm just quantum probabilistic.

p.H
Posts: 1969
Joined: 2017-09-17 07:12
Been thanked: 9 times

Re: Migration from Windows10 to Debian10

#10 Post by p.H »

e151280 wrote: 2021-07-26 18:33 the motherboard's Intel Matrix RAID controller is seen as a "fake raid"
FakeRAID can be troublesome with Linux and does not provide real advantage over software RAID.
If I were you, I would save the data and remove all SATA RAID arrays. If you want RAID, you can use Linux software RAID or a real hardware RAID controller.
wizard10000 wrote: 2021-07-26 18:53 RAID0 will double your seek time
Can you explain how RAID 0 doubles seek time ?
wizard10000 wrote: 2021-07-27 12:31 sector density is a much better indicator of disk throughput than rotational speed will ever be
Sequential speed is proportional to rotational speed.
Sequential speed is proportional to the linear component of recording density. For example, SMR increases only the radial component of recording density, so the increase of overall recording density does not increase sequential speed.

User avatar
wizard10000
Posts: 237
Joined: 2019-04-16 23:15
Location: southeastern us
Has thanked: 28 times
Been thanked: 10 times

Re: Migration from Windows10 to Debian10

#11 Post by wizard10000 »

p.H wrote: 2021-07-29 20:22Can you explain how RAID 0 doubles seek time ?
"Double" may be a bit of an oversimplification but seek time on a RAID0 array will never match the speed of a single drive of same make and model because system overhead processing multiple read/write requests.
p.H wrote: 2021-07-29 20:22Sequential speed is proportional to rotational speed.
Sequential speed is proportional to the linear component of recording density. For example, SMR increases only the radial component of recording density, so the increase of overall recording density does not increase sequential speed.
True. Still simplifying, the more sectors per track the faster sequential read/writes will be. Anecdotal evidence, but I bought a rather spendy wide SCSI RAID5 setup with a real Adaptec RAID controller. Three 9G 10k rpm enterprise drives.

Needed more storage so I bought a 40GB 7200 rpm Hitachi spindle that outperformed the hell out of that spendy RAID array. I went back to single spindles and sold that RAID controller and the three drives for about half what I paid for them :)
we see things not as they are, but as we are.
-- anais nin

CwF
Posts: 1189
Joined: 2018-06-20 15:16
Has thanked: 2 times
Been thanked: 6 times

Re: Migration from Windows10 to Debian10

#12 Post by CwF »

wizard10000 wrote: 2021-07-30 02:43 40GB 7200 rpm Hitachi spindle that outperformed the hell out of that spendy RAID array.
It didn't. I built many and there is a timeline your forgetting. It was a stepping stone leapfrog situation. A current IDE never outperformed a current scsi. When IDE's were at 3-5MB/s SCSI was slightly higher with a 20MB/s bus to pool multiples. Then the single rate eclipsed the 20, there was 40, and 80, and so on. The viability of any stage was a few years. My 12 disc arrays (2x6 80pins) ran for a decade and could saturate the then more rare gigabit (110MB/s+). It would give a modern generic sata spinner a run, and the modern version wouldn't touch seek times or iops.

The method now should be simply use all the available ports.

pendrachken
Posts: 1394
Joined: 2007-03-04 21:10
Location: U.S.A. - WI.

Re: Migration from Windows10 to Debian10

#13 Post by pendrachken »

e151280 wrote: 2021-07-26 18:33 Dear fellow Debian forum users,

thank you for the "ntfs-3g" clue for maintaining access to my multimedia storage volume. Btw, I see 2 recommendations to NOT create a ssd raid boot volume. I was thinking that while SSD is fast, 2 SSD drives in RAID 0 would be even faster (simultaneous reads). What danger am I not seeing? I also have a battery backed Adaptec 6805 RAID card from an old server that I can add in, if the motherboard's Intel Matrix RAID controller is seen as a "fake raid".

Please enlighten me... I am ALWAYS willing to learn.
RAID0 is hot garbage for holding things that you don't want to lose. Like your OS.

RAID0 is one drive failure away from your data being garbage on the rest of the drives. And the more drives you stripe across the more failure points you have. For example:

Code: Select all

Lets just say, for the sake of argument, that you got drives from a bad run of SSDs that run fine right up to instant death. The failure rate works out to 45% chance for a drive to fail in one year. 

Then you have a RAID0 array striped across two drives. Simple statistics says you add probabilities when you have an "OR" operation - such as drive one OR drive two chance of failure. 
So we add the probabilities of failure - 45 + 45 = 90% probability of at least one of the drives failing in the first year of use. 
Probabilities for three+ drives in the array would indicate that it's almost a certainty that at least one drive in the array would fail, unless you are lucky enough that you should be buying lottery tickets for every drawing because you always get the jackpot.

With RAID0 that means your OS is dead, any customizations you did are dead, and any data you had on the OS drive is gone. Unless you made backups that can be restored. Not worth it at all. 

Now are the numbers as bad as that? No, but why double or triple your probabilities of drive failure for no reason?

Boot speeds are more than fast enough on SSD. Unless you have some strange edge case hardware that has detection problems in the kernel or something, you shouldn't be looking at more than 20 seconds to boot.... at the long side. Plus you don't need to shut down / reboot Linux as much as you do Windows. Pretty much the only thing that ever needs a reboot for is kernel security updates. And there are ways around even that if you care enough or run a production server that just can't go down ever.
fortune -o
Your love life will be... interesting.
:twisted: How did it know?

The U.S. uses the metric system too, we have tenths, hundredths and thousandths of inches :-P

CwF
Posts: 1189
Joined: 2018-06-20 15:16
Has thanked: 2 times
Been thanked: 6 times

Re: Migration from Windows10 to Debian10

#14 Post by CwF »

pendrachken wrote: 2021-07-30 04:03 you shouldn't be looking at more than 20 seconds to boot....
Ha! I'm lucky to get a post within 20 seconds!

But you're right,
CwF wrote: 2021-07-30 03:35 The method now should be simply use all the available ports.
I should have finished that with " for independent ssd's and spinners, one of which is a boot ssd."

User avatar
wizard10000
Posts: 237
Joined: 2019-04-16 23:15
Location: southeastern us
Has thanked: 28 times
Been thanked: 10 times

Re: Migration from Windows10 to Debian10

#15 Post by wizard10000 »

CwF wrote: 2021-07-30 03:35It didn't.
Yeah, it actually did :)

ATA-4 has more bandwidth available than wide SCSI does. Plus, the IDE drive had about 4x more sectors per track than the SCSI drives did.
we see things not as they are, but as we are.
-- anais nin

CwF
Posts: 1189
Joined: 2018-06-20 15:16
Has thanked: 2 times
Been thanked: 6 times

Re: Migration from Windows10 to Debian10

#16 Post by CwF »

wizard10000 wrote: 2021-07-30 04:28 Yeah, it actually did :)
Sure, like I said, leapfrog every few years. I played the same game back then, and the chase is why I'm more KISS now and somewhat resistant to shiny things!
Actually my preference to run multiple independent disk did come out of the scsi arrays. Like you say, each new generation changed the assumptions. A disk/ssd on it's own port doing a particular job is much more effective than everything coming off a single array. Speaking of 'few' user workstations here not 1000 user servers. So, a disk per single task - boot/OS disk, a worm-like storage disk, a program disk (/homes), maybe a game disk, scratch disk, whatever...that last one I've moved from ssd to ram since that has it's own change in nature - cheap and plentiful now.

User avatar
wizard10000
Posts: 237
Joined: 2019-04-16 23:15
Location: southeastern us
Has thanked: 28 times
Been thanked: 10 times

Re: Migration from Windows10 to Debian10

#17 Post by wizard10000 »

CwF wrote: 2021-07-30 12:54Sure, like I said, leapfrog every few years. I played the same game back then, and the chase is why I'm more KISS now and somewhat resistant to shiny things!
Guess I misunderstood your post - sorry :)

Yup, simpler is better. What I did with that wide SCSI array is something I like to call intellectual masturbation. You get a happy ending when you get the thing running and then learn later it was a stupid idea :mrgreen:

I have other examples; I built this quad-boot monstrosity once running Win 3.11, a Win95 beta, OS/2 Warp and RedHat Linux. The whole mess was held together by a boot manager called System Commander and I spent more time fixing the damn thing than I did using it. I was pretty darned proud of myself for tweaking OS/2 to give me > 700KB RAM in a DOS box. Didn't occur to me until much later that at the time no DOS application could access more than 640K.

In another stroke of brilliance I decided that putting a Win 3.11 swapfile on a ramdisk was a Great Idea. Pretty sure I only had 4MB of RAM in that machine.

I think most geeks go through a phase like this - mine lasted for about five years :)

cheers -
we see things not as they are, but as we are.
-- anais nin

CwF
Posts: 1189
Joined: 2018-06-20 15:16
Has thanked: 2 times
Been thanked: 6 times

Re: Migration from Windows10 to Debian10

#18 Post by CwF »

wizard10000 wrote: 2021-07-30 13:33 swapfile on a ramdisk
This idea has come back around, zramswap! Details can be debated...
I recycled a 80 pin cage and vfio passed a scsi controller for a multiboot/hotswap VM. Fun!
-personally, life is a waste of time - waste it well.

p.H
Posts: 1969
Joined: 2017-09-17 07:12
Been thanked: 9 times

Re: Migration from Windows10 to Debian10

#19 Post by p.H »

wizard10000 wrote: 2021-07-30 02:43 seek time on a RAID0 array will never match the speed of a single drive of same make and model because system overhead processing multiple read/write requests.
Can you elaborate how system overhead is significant in front of disk latency ?
wizard10000 wrote: 2021-07-30 04:28 ATA-4 has more bandwidth available than wide SCSI does
Wide SCSI was obsolete at the time ATA-4 came out.
pendrachken wrote: 2021-07-30 04:03 The failure rate works out to 45% chance for a drive to fail in one year.
This is ridiculously high.
pendrachken wrote: 2021-07-30 04:03 Then you have a RAID0 array striped across two drives. Simple statistics says you add probabilities when you have an "OR" operation - such as drive one OR drive two chance of failure.
This is an appoximation which is valid only when the probability is much less than 1 (p << 1). It is not valid for p = 0.45.
The actual probability is p(2-p).
pendrachken wrote: 2021-07-30 04:03 So we add the probabilities of failure - 45 + 45 = 90% probability of at least one of the drives failing in the first year of use.
No, probability of failure for p=0.45 is p(2-p) = 0.7.
pendrachken wrote: 2021-07-30 04:03 Probabilities for three+ drives in the array would indicate that it's almost a certainty that at least one drive in the array would fail
With your approximation, probability would raise above 1, which is impossible. Doesn't that ring a bell about its validity ?

pendrachken
Posts: 1394
Joined: 2007-03-04 21:10
Location: U.S.A. - WI.

Re: Migration from Windows10 to Debian10

#20 Post by pendrachken »

This is ridiculously high.
It's almost, ALMOST like it's an over simplified example... And nice cherry picking by completely ignoring the prior sentence too.
This is an appoximation which is valid only when the probability is much less than 1 (p << 1). It is not valid for p = 0.45.
The actual probability is p(2-p).

No, probability of failure for p=0.45 is p(2-p) = 0.7.
No, that would be the assumption that the failure is not mutually exclusive, when any single drive failure will kill the array making it exclusive. It also is about exclusivity, not how high or low the probability is for any given event. By your reasoning flipping a coin wouldn't ever reach p=1 since you aren't just adding p(A)=0.5 + p(B)=0.5 to get P(A) OR p(B)=1

You would be correct IF you assume the second disk would be be re-used in a rebuilt array. Under the mutually exclusive rules where the array dies and all data is lost, and the other disk is no longer in use it is simply p(A) + p(B).
You could also go back and calculate the probability of BOTH disks going bad, but there is no reason to. In this example the damage is done when one dies since it is in RAID0.
With your approximation, probability would raise above 1, which is impossible. Doesn't that ring a bell about its validity ?
Only that it's a certainty, barring stupendous luck, that it would fail in that time frame. To get a probability under 1 for a certain event in a time frame you would have to calculate the probabilities in a more constrained envelope. If you reach a probability of 1 in six months, that doesn't mean the outcome doesn't happen or changes if you are looking at the time frame of 1 second through one year.
fortune -o
Your love life will be... interesting.
:twisted: How did it know?

The U.S. uses the metric system too, we have tenths, hundredths and thousandths of inches :-P

Post Reply