sawsedge,
I did manage to clone systems with different geometries but had not investigated the process deep enough to make any useful contribution. The problem is of course hard disk specific. I also remember the cloned disks did have a stability problem of some sort.
All in all I would move with time and forget the old disks. AFAIK all hard disks sold nowadays are standardised in LBA mode with 255 heads and 63 sectors.
Yes I will do that as soon as I get the new board/cpu etc.
I had noticed on boot that the 20g is always reported as CHS, the rest of the drives as LBA. Thinking back on it, I believe I got that drive when I had a Celeron 366, so it is quite old.
Thanks again.
Little background...I had a disk that was showing signs of failing in my laptop. I purchased a new hard drive and usb enclosure to do a clone. Both gparted and parted magic would not even see the fact that there was another hard drive connected via USB. I gave the g4l live cd a shot. It saw the hard drive but would fail to copy.
Then I tried a knoppix live cd and all worked well. Then I used gparted live cd to do the partition resizing. All went perfect.
Just throught I'd throw that out there for people trying to clone a drive connected via USB. You may have to use Knoppix instead of gparted live cd and then just follow the directions listed in the first post.
drsatch,
I rather think your problem is hardware related.
It is a common fault that if a USB hard disk isn't properly powered then it may fail to be picked up by some operating systems.
The maximum power supply to a USB port is 0.5 Amp. Just about every 2.5" hard disk above 20Gb capacity would be using 0.5 to 1 Amp consumption range. A properly kitted out external 2.5" hard disk enclosure would have a twin head USB cable supplied for drawing power supply from two USB ports.
If you are using a twin head cable and the distro is a recent one then your USB hard disk should be seen by every Linux.
If you use a single head USB cable then the hard disk read/write operations are simply "unreliable". Sometimes it works but sometimes it doesn't. This is not an operational problem.
Can you confirm what cable you were using?
Also the best detection is at the terminal using command in root
Code:fdisk -l
You're exactly right. It was a twin head. I was using 1. Never thought about the power requirements. Just thought it was strange that parted magic and gparted iso's wouldn't see it, but the knoppix one did. I did use the command line to find the disks also.
It's exactly as you said...sometimes it works using one lead, sometimes it doesn't.
In my case, it did. All is well in my world.
drsatch,
Thanks for the confirmation.
I just want to thank you for talking to us, like the children we are. Not all of us was born with this embedded in our brains. About 2 minutes ago I completed my first XP migration. Your instructions were simple and to the point. The only issue I had was that the code you gave , ( dd if=/dev/hda of=/dev/sda bs=32256) had to be changed to (dd if=/dev/sda of=/dev/sdb bs=32256) because thats the way terminal had designated it. Could you show a me a code, lets say , to single out one partition of a duel boot, with XP and linux, on a single drive. I think if I seen this it would better help me understand, this process . Thanks again AD johnson:cool:
AD Johnson,
The safest way to clone a MS system, especially Vista, is to mirror a complete hard disk.
Thereafter if you want to change the partition arrangement do it with the Parted Magic or Gparted Live CD. This way you start with a workable system first, test it and then start messing around with the partition table. For Vista it is best to use its own resizer function within the "disk management" program. Parted magic and Gparted are alright for Xp.
To clone a partition just use the device name of a partition. In Linux the first partition of disk sda is sda1, the 3rd partition of disk sdb is sdb3. Therefore is you want to clone sda1 as a source to sdb3 as a target the instruction is just
You need to do extra work to change some system parameter if you migrate a MS system or any operating system to a partition with a number different from its original. That is why I recommend cloning the whole disk first and alter its partition table later. Also you need to make sdb3 exactly in size, in exactly the same number of cylinders, as sda1 in order to be error free.Code:dd if=/dev/sda1 of=/dev/sdb3 bs=32256
In general cloning a partition without an operating system (data only) is recommended for the above. However you might find creating sdb3 first, format it and then copy the content of sda1 into it is easier as you can have sdb3 in any size you wish.
Cloning with dd has the advantage of copying the boot sector which is not touched by programs copying the filing system.
If you want to know how each partition is called in Linux use a root console command
Lastly it was not my intention to make my readers feel like children. However the operation is very simple but the theory behind is quite complicated. My intention was to make aware of the principles behind. To migrate an operating system by any standard is a big deal so there is no harm in stating each step clearly.Code:fdisk -l
I Thank you for your quick response. It has shed a lot of light on the subject. Right now I'm trying the ( tar ) method in migrating .elyssa mint 5 to an external hard drive,just to see how it works. Again I thank you, I hope I did not offend you with my children statement, But when I go to some forums, they talk as if everybody's been doing this all they're lives. Nice to listen to someone who talks to us in a way we understand. Thanks
First, this is a most helpful how-to. Much thanks to saikee and all who've contributed to this thread.
I've run into a temporary glitch trying to use this method and have a question:
Can the destination drive be booted from while still in the external enclosure and using a different interface (SATA vs. IDE) than the original drive? I can't seem to boot from the destination drive that I've copied the data to.
Here are the steps I've taken:
1) System: Dell Latitude D420, 60 GB IDE running XP Pro, external USB drive enclosure housing WD 74 GB SATA drive. Using Xubuntu livecd.
2) Fdisk output:
ubuntu@ubuntu:~$ sudo fdisk -l
Disk /dev/sda: 60.0 GB, 60011642880 bytes
255 heads, 63 sectors/track, 7296 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x41ab2316
Device Boot Start End Blocks Id System
/dev/sda1 1 7 56196 de Dell Utility
/dev/sda2 * 8 7294 58532827+ 7 HPFS/NTFS
Disk /dev/sdb: 74.3 GB, 74355769344 bytes
255 heads, 63 sectors/track, 9039 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x39d1cc9a
Device Boot Start End Blocks Id System
3) dd command and output:
ubuntu@ubuntu:~$ sudo dd if=/dev/sda of=/dev/sdb bs=32256 conv=noerror
dd: reading `/dev/sda': Input/output error
14401+1 records in
14401+1 records out
464535552 bytes (465 MB) copied, 184.913 s, 2.5 MB/s
Displays i/o error three more times and continues processing.
I added the conv=noerror because the dd process would die upon hitting the error, where the noerror kept things moving.
I also found a useful way to monitor the dd progress at another site (forgot where, but I'll find/link-to it ASAP). In another terminal enter:
$ sudo watch -n30 -- pkill -USR1 ^dd$
The above will display the progress in the original terminal which the dd command was executed.
4) Last statement in dd progress:
60011581440 bytes (60 GB) copied, 6167.66 s, 9.7 MB/s
5) Fdisk output:
ubuntu@ubuntu:~$ sudo fdisk -l
Disk /dev/sda: 60.0 GB, 60011642880 bytes
255 heads, 63 sectors/track, 7296 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x41ab2316
Device Boot Start End Blocks Id System
/dev/sda1 1 7 56196 de Dell Utility
/dev/sda2 * 8 7294 58532827+ 7 HPFS/NTFS
Disk /dev/sdb: 74.3 GB, 74355769344 bytes
255 heads, 63 sectors/track, 9039 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x41ab2316
Device Boot Start End Blocks Id System
/dev/sdb1 1 7 56196 de Dell Utility
/dev/sdb2 * 8 7294 58532827+ 7 HPFS/NTFS
ubuntu@ubuntu:~$
Looks good, but when I try to reboot the laptop via the external enclosed drive, BIOS says: A disk read error occurred, ctrl alt del.
XP comp mgmnt says: e:\ is not accessible. The file or directory is corrupted or unreadable.
Sorry if too much info. Any help or point in the right direction is appreciated.
--shortz
shortz,
Welcome to JustLinux.
Xp or any MS Windows system does not support booting from an external device, except eSata that is factory-fitted at the rear panel or a specially hacked version.
Also it is not wise to have two Xp systems, with identical system files in any booting operation. It is much safer to remove one of the two disks and arrange it in exactly the same hardware condition as such information would have been embedded into the system during installation. This is to say it will not boot from e: if it was originally installed in c:.
As far as I can tell your original disk may have some corruption resulting reading difficulties with some sectors. The rest appears to have been successfully cloned.
Depending on the age of the laptop your old disk may be a Pata and the cloned Sata may not be physically fitted into its internal hard disk position.
If you intend to replace the hard disk get one that is of the same type. You can clone the 3rd disk from either of the two disks you have got.
saikee,
Thanks for the quick and thorough response!
I did not know that XP could not boot from an external drive, with the exception of eSATA, etc. Good to know.
What I termed IDE (incorrectly) is likely PATA. Yes, the external drive is too large to fit into the laptop. My original plan was to backup XP, resize the NTFS partition smaller and install another OS for multiboot. In case I messed things up or wanted to return the system to original setup I hoped I could use the backup and dd from external drive to internal drive.
I will obtain a matching disk (interface, enclosure, etc) and try the backup and replace method next go round.
Again, thanks for all your help as well as the contribution from others on this thread!
--shortz
The Dell Latitude D420 is a laptop. All laptops use 2.5" hard disk as oppose to the 3.5" size hard disk used in the desktop.
I would advise you to remove the hard disk of the laptop and check the connection. The IDE or Pata has two rows of pins whereas a Sata doesn't, otherwise the two look exactly the same externally. It may be possible for you to check the original spec to find out if it is a Pata or Sata hard disk.
First off, wanted to give a big thank you to everyone here for comments and suggestions, and would like to get thoughts from those there on my situation.
I need to re-image drives quite often, and for my purposes I need to do the WHOLE drive. (All partitions, including boot record.) (If you REALLY need to know why, PM me. :D)
Because my images are stores on a CIFS mountpoint, speed and size is an issue. (i.e. I CAN do a dd if=/dev/sda of=/mnt/cifs/image.out but it would take quite longer than necessary.) So, looking at alternatives, 'partimage' will copy ''just what's filled'' on the drive. However, it only does partitions and not the whole drive.
So my proposed solution (working on now) is this, in high-level, using a custom Live Linux CD that will boot up into Text mode and run the following in a script:
1. Mount the network share.
2. Write the new partition table - sfdisk /dev/sda < table.txt
3. (Resync the table?)
4. Use partimage to restore all the partitions.
5. Use dd to write the first 1k (1024) bytes of the drive, to make sure the full MBR is there.
Again, this is high level, but does that seem like it has any huge holes in it? And, is there an easy way (other than rebooting) to resync the table? I saw someone mention "partprobe" but not sure if that'll do anything it needs to.
Thanks in advance all,
Mike.
Coyttl,
Welcome to Justlinux!
I suppose your backup system is always bootle-necked by the network speed.
You idea of breaking up the hard disk image into parts should work as long as you make sure the first sector, which contains the partition table, is saved as part of the procedure.
I often back up a large hard disk, say 1 TB large, but only interested in the first partition. The quickest way is to "dd" the first sector out and put it on a say a smaller disk. The transferred partition table makes the smaller hard disk a outcast because its partition table points to higher number of cylinders the disk hasn't got and most partitioning software would not touch it. The fdisk program in Linux can be used to delete the unwanted partitions making the smaller disk "legal" again.
The partition table is the guide for program like dd to carry out cloning as each partition's boundary limit is adequately defined.
Thus if you restore the first sector, only 512 bytes by command
You can restore any partition in piece meal fashion.Code:dd if=/dev/sda of=/mnt/sda_1st_sector bs=512 count=1
You do need to make sure the partition table isn't altered.
Also if you use logical partitions and LVM it will be necessary for you to adhere to the partition layout rigidly as these partitions addresses are hard-coded between partitions and not inside the first 512 bytes.
Right now that's true, and as soon as I get everything 're-imaged' to a new, more useable fashion, the net's going to be upgraded to a fiber backbone to speed up data transfers. :) But that's a little ways off...Quote:
Originally Posted by saikee
Yeah, I got that.. In my case, since we're doing software testing on many different flavors of OSes (mostly Windows), sometimes I need the entire disk (read: All partitions and data. Empty space I don't care about) since an OS on a machine from a manufacturer may have a "backup" or "data" partition. Unfortunately, I gotta save those as well.Quote:
Originally Posted by saikee
Ah, okay! So I can use these commands together:[CODE]sfdisk < partitions.txtQuote:
Originally Posted by saikee
dd if=/dev/sda of=/mnt/sda_1st_sector bs=512 count=1[/COUNT]..instead of saving 1024 bytes, I only need the first 512... Right?
I'm not two worried about ever having to change partitions - since the main point of my imaging is to 'restore an operating system back to it's fresh standing' without the software we're testing on it, the drives themselves will never change.Quote:
Originally Posted by saikee
My main concern for this whole process I mad-scientist-formulated was because of MBR issues and my old way of imaging. Newer OSes (Win08, Vista) didn't like my previous imaging way. Add newer AHCI controllers to the mix, and I'm dead in the water on newer machines now. Hence, my foray into handling it this way. :)
thanks for the pointers!
Mike.
Not if sfdisk is at least halfway intelligent. (Though I don't know whether it is.) If it's at least halfway intelligent, then it will send the proper ioctl to /dev/sda for you, so that the kernel re-reads the (new) partition table off the disk. The only way this should ever fail is if any other partitions on the disk are currently mounted.Quote:
Originally Posted by Coyttl
The partprobe program does send this same ioctl -- but if sfdisk is doing it for you, you don't need partprobe. ;)
Thanks for that. I'll test it later today and see what happens. I have more than a few systems I can (potentially) destroy. :)Quote:
Originally Posted by bwkaz
I'll post what I find (for anyone else's reference..)
Very useful topic - so far! I've succesfully migrated my XP installation from an 80GB IDE to a 250 GB SATA. Due to errors on the disk I did have to use ddrescue rather than dd but I got there eventually and it all boots up just fine.
However, resizing the cloned partition has not been so easy. the new disk shows up as having one NTFS partition of 80GB and the rest as Unallocated space. I've tried using Gparted and the Parted Magic Live CD but nothing seems to allow me to expand the existing partition to take advantage of the additional space on the new drive. I am allowed to set up the new partition size but when I "apply" the operation I get a rather unhelpful message telling me that the operation couldn't be performed.
From the description of the migration process detailed at the start of the thread there is no indication that this operation should present any problems but all my attempts to accomplish it have met with failure.
Am I missing something obvious?
TarsTarkas,
Welcome to Justlinux!
Xp from a 80Gb disk source to a 250Hb disk should be plain sailing. The expansion of 80Gb to 250 should take about a minute or two as only the right hand side boundary is changed and the filing system can remain unmoved. I believe the operation only adjusts the end point of the hard disk and generates the indexing system for the NTFS filing system between 80 and 250Gb space.
I would suggest you defrag the Xp partition first, use Gaprted or Parted Magic to drag the right boundary from 80Gb to 250Gb and give the green light to go ahead.
If you still have a problem post the disk management details here (i.e. telling us how manay partitions and where they are primary or logical.)
The resizing may not work with encrypted filing system though.
Hi Saikee!
Many thanks for swift response. I've subsequently defragged my disk (three times to make sure there are no red bits left) and retried Parted Magic but I still get the same error. Here's the details:
GParted 0.4.1
Libparted 1.8.8
Grow /dev/sda1 from 76.68 GiB to 116.45 GiB 00:00:00 ( ERROR )
calibrate /dev/sda1 00:00:00 ( SUCCESS )
path: /dev/sda1
start: 63
end: 160810649
size: 160810587 (76.68 GiB)
calculate new size and position of /dev/sda1 00:00:00 ( SUCCESS )
requested start: 63
requested end: 244204064
requested size: 244204002 (116.45 GiB)
new start: 63
new end: 244204064
new size: 244204002 (116.45 GiB)
check file system on /dev/sda1 for errors and (if possible) fix them 00:00:00 ( ERROR )
ntfsresize -P -i -f -v /dev/sda1
ntfsresize v2.0.0 (libntfs 10:0:0)
Device name : /dev/sda1
NTFS volume version: 3.1
Cluster size : 4096 bytes
Current volume size: 82335019520 bytes (82336 MB)
Current device size: 82335020544 bytes (82336 MB)
Checking for bad sectors ...
Bad cluster: 0x50829 - 0x50829 (1)
Bad cluster: 0xdd0f92 - 0xdd0f92 (1)
ERROR: This software has detected that the disk has at least 2 bad sectors.
************************************************** **************************
* WARNING: The disk has bad sector. This means physical damage on the disk *
* surface caused by deterioration, manufacturing faults or other reason. *
* The reliability of the disk may stay stable or degrade fast. We suggest *
* making a full backup urgently by running 'ntfsclone --rescue ...' then *
* run 'chkdsk /f /r' on Windows and rebooot it TWICE! Then you can resize *
* NTFS safely by additionally using the --bad-sectors option of ntfsresize.*
************************************************** **************************
========================================
I followed the advice and ran "chkdsk /f /r", rebooted twice then reattempted to resize the partition but the bad-sectors are still reported. The information above asks me to run ntfsresize to resize the file system but I'm unsure how I can do this without first increasing the size of the partition ...
... or are they the same thing?
If you have bad sectors in the hard disk then these can be corrupted patches which are not readable or writable. This is a hardware problem and can only be the case if you are using a used 250Gb hard disk.
A brand new disk should not have bad sectors. I certainly have not run into one of such disks myself.
That's a little disturbing since the 250Gb drive was purchased from PC World a month ago (today) with no indication that it was either second hand or refurbished. I bought it to replace the 80Gb drive which did have bad sectors and was generally getting a bit creaky.Quote:
Originally Posted by saikee
Now if you'll pardon my ignorance as to how dd and dd_rescue actually work, but if either command is just making a "dumb" copy of the original drive is it possible the bad sectors have been copied "intact" from the original drive. I believe that bad sectors are marked as "no-go" areas and I'm guessing that the cloning process simply copied over the marker flags. If this is case would you know how to "unmark" them.
Failing that, it'll have to be a trip back to PC World. What a pain! Glad I held on to my receipt.
Many thanks for all your help, however it turns out. 'Tis much appreciated!
dd has an option to proceed after encountering an error. My guess is dd_rescue may do a bit more.
If you copy from a source with bad sectors the defective areas cannot be read and it would not be copied but just left blank on the target disk. The blank area should have no problem to be read and written. Only when the file is being used then the missing information would be shown up.
Technically the bad sectors cannot be transferred and that is certainly my experience when cloning dying disks.
If the disk is bad the defective areas will be reported by any cloning software. You can try one from Windoze like Ghost or Acronis. dd just copies the binary pattern of "1" and "0". It is a pretty basic operation. The good thing about it is the resident operating system is not used and its type, health and defects are immaterial to the cloning operation.
The dd method is solid and simple and resolved my first question of whether or not I could migrate my system as is to a larger hard drive; two quhow to migrate linux system to a new driveestions still:
1. I have been reading about increasing speed and optimizing ubuntu for laptops and wanted to change the default ext3 file system to xfs or or another faster file system. Is this possible to do before or after the dd process?
2. Is it possible to use this method to create a bootable sd card of a current ubuntu setup? how to migrate linux system to a new drive
Dylanspiritwind,
Welcome toi Justlinux!
In answering your questions
(1) All information are stored in binary bits of either "1" or "0" on a hard disk so the filing system and the operating system is immaterial. In your case you can do either because one of them is your backup so you can do whatever you want even damage the second disk without any consequence to your original system. Thus it may be easier to have two ext3 disks before proceeding to turn one of them into the xfs filing system.
(2) The answer is yes but it is a lot easier and quicker to use Ubuntu to do it in terminal mode. The details are available in this thread. Forgive me if I do not repeat them here. Just follow the suggestion to put Grub into a pen drive. If you don't want to read it I believe if you insert the sd card and it is recognised as device sdb1 then the Bash terminal command
should be able to achieve the same but I am not sure if it wipes away the existing data in the SD card.Code:grubinstall /dev/sdb1
Migrating Linux is easier than cloning it and you can do it with itself but the safe way is always booting up a Live CD leaving the original system dormant so that its system files are not in active use.
Key steps are (let us know if you need the missing commands)
(a) Do it one partition at a time.
(b) Just create and format another target partition of any size and any filing type (same filing type is always the best).
(c) Mount both source and target partition. (assuming the source is /dev/sda3 mounted on /mnt/sda3 and target /dev/sdc3 mounted on /mnt/sdc3)
(d) Change directory to the source /mnt/sda3 (not mandatory but the command below is written assuming you do it) and then copy the entire filing system across by command
(e) Do whatever necessary to update /etc/fstab to point to the new target partition by keeping or removing the source partition.Code:tar cf - . | (cd /mnt/sdc3; tar xf -)
(f) As a rule duplicating a file system using a "copying" command does not copy the boot sector and so the boot loader must be updated separately. dd is the only command that includes the boot sector.
Thank you for your quick reply!
I did read the link, but it is a bit over my head, are they doing that in grub?
So my sd card comes up (in diskpartitioner) as sda2 but has 2 partitions on it called sdb1 and sdb2, I am not concerned about any information on it so would I use the command grubinstall /dev/sda2? Being that I have only 1 sd slot to boot my live ubuntu from and no cd/dvd Im not sure I can use the dd command from outside my running system.
My system is basically as follows:
eee 900ha (thus no optical drive)
160gb hd that is partitioned to dual boot (still have the xp on there even though I never use it) and partitioned:
Sda1:[/B] ntsf/windows
Sda2: sd card
sda3: says unknown, 40 mb?
sda4: extended:
I believe those are my 4 primary partitions?
under sda4 it shows: sda5: ntfs, 235mb, not sure what this is for?
sda6: I think this is my primary linux partition, it has 7.11gb used (will that fit on a 8gb sd card?
sda7: linux swap
sda8: fat 32: this is my partition that holds all my documents/music/videos/etc.
So after installing grub onto the sd card,
(a). which partitions do I need to copy
(b). Im not sure I understand the other steps needed here, do I need to do any formatting to the sd card?
(c,d,f): This is a cf command and is tarred, would it be easier to use the dd command instead here?
(e). Im not sure I understand this.
After using Linux for a year, I guess im still a NooB,
My goal here again: I just want a bootable backup of my eeebuntu os (I have a bootable copy of the original, just want one with all the tweaks I've made), I could settle for a dd backup that I could restore but it would be re-assuring to be able to boot into my system instead of the original version.
After some searching I found remastersys to create a custom .iso image, could I simply do that then boot into windows and use unetbootin to creat that on a flash drive? I think I will try that now:)
If this same result would be possible using the dd command it would be nice, the remastersys does allow to "reinstall" the system just like a live cd when booting from it which is even more ideal for a beginner like me.
Thanks in advance!
Your sda2 should not be a SD card.
The information suggests that you have two disks; one sda (a hard disk) and a SD card called sdb.
I think I have been confused by you as I thought you just want a bootable SD card containing Ubuntu settings. To me that can only be the boot loader Grub and nothing else. If you want the SD card to contain a full working Ubuntu then it is something else.
You will need to check the Ubuntu forum on making a SD card version as it doesn't appear to be available from the general release the last time I tried it.
In your case you would be better to resize the partition to squeeze some hard disk space to create say sda9 and put the second Ubuntu there. The second can share the swap partition and just one partition will suffice. You can boot it as another choice in the Grub menu.
The problem with booting from an SD card is you must nominate the sd card as the first bootable disk before the PC Bios drops the booting to the hard disk. Therefore the SD card will take the sda device name and the internal hard disk is relegated to device sdb. You may have to change the internal settings when using the SD card and then revert back to the original setting if booting the internal hard disk first. You end up with more trouble than its worth.
Putting two Ubuntu into two partitions simplifies everything.
You were so right, I am a bit too *****ious for my own good,
The theory worked great, Using Remastersys to create the iso image then unetbootin to make a bootable sd card, but my sda 2 was not the sd card....
So now my computer boots with the unetbootin version (I presume off of sda2, as the sd card is empty) but I have lost my ability to choose between the original version on sda6 and my windows on sda1. I think this method should work once you have the iso (I used unetbootin to installl eeebuntu in the first place, I dont see how it could be any different), just dont make the same mistake I did, I hope I can salvage my original grub menu (or boot menu?).
I will look for help in another forum that is more on this topic, thanks!
I just got done replacing the disk in my Dell Inspiron laptop. Since I dual boot between XP and Linux the original 80GB drive was becoming cramped.
Dell said a 120GB drive was the largest I could use. My local computer store was willing to let me try a 320GB drive and return it if it didn't work.
The basic clone procedure in the first post worked fine. Took about 90 minutes and the new disk booted to both XP and Linux.
The problem was that I was unable to use the additional space.
I won't detail all of the things I tried before the epiphany.
It's important to remember that laptops from vendors have special partitions for various purposes. I used Partition Magic when I made room for Linux when I first got the Laptop. I also created a "fat32" partition so that I could share files between the two OS's.
While setting this up I used up the primary partitions and created an extended partition for Linux and Swap. It wouldn't let me resize the extended partition.
This is what created the problem stated above. I didn't want to disturb the factory partitions. The main goal was to not have to reinstall windows, just give it more space.
Part of the reason for wanting extra space was to be able to install more than one Linux distro. I was willing to blow away Linux and reinstall. Installing Debian and getting 90% of what I need takes a little over an hour. With the recent release of Lenny and the fact the you can get install disks with your choice of desktop, XFCE, GNOME, and KDE, I wanted to try the XFCE version.
Of course after deleting the extended partition the laptop wouldn't boot anymore. Grub would fail due to the menu.lst file now missing. I had total confidence that the Linux install would find XP and set-up Grub appropriately. It did!
Windows did run CHKDSK when it was first booted due to the new larger space.
Everything is working great. I just have to change the sources to "testing" and do a distribution upgrade.
This turned out to be a great learning experience. The information in this thread was helpful. I wanted to add my lesson of what might happen if the disk you're trying to clone has used up it's primary partitions and also has an extended partition. It adds complication and some additional planning is needed. The beauty is that the original disk was always safe. I could mess up the new drive without fear that I could start over.
Thanks,
teeitup,
Glad you find the procedure works for you.
To expand an extended partition is more complicated. If the OSes are Linux you don't need to clone them. Just use the tar command and then restore the boot loader one by one.
It is a good practice to put the extended partition at the rear end of a hard disk. This way you can expand it easily.
I did have the extended partition at the end. I thought it was going to be a no brainer. I was disappointed and a little concerned that my original plan didn't work. The idea of 320GB of disk space had grown on me and I didn't want return the disk.
I found it interesting that the extra space was identified as "unusable" and not "unallocated". This was true in GParted and cfdisk.
Thanks,
teeitup,
The hard disk has a standard that every OS uses. One of the standard features is the 4 primary partitions. Basically you can have only more partition by giving up one primary and turn it into an extended partition.
Inside the extended partition, which is just a boundary, you can have 11 logical partitions (pre-2.6.28 kernel) or 59 logical partitions (post 2.6.28 kernel). The logical partitions must be consecutive because the ith partition carries the address for the i+1 partition. If you delete say the 7th partition the space is dead but the Linux will shift the 8th partition and above up by one space so that the old 8th becomes the new 7th, old 9th become the new 8th etc to maintain the consecutive order, even without telling you!
If you have space outside the above convention no operating system uses it. That is all.
Therefore in your case if you have dead space after the end of the logical partitions you can delete a couple of them and move the boundary of the extended partition to take up the dead space.
Dead space can be easily absorbed by primary partitions but it is more tricky with logical partitions because of the convention. There is a reason for it. If you study the convention you will agree that is the optimum way.
Saikee,
Great thread as always, though I could use your help. I have a drive that is failing on me. My Fedora partition no longer boots and fsck ran overnight without completing. :/
So, I recently purchased a 500GB drive and planned on cloning my main Linux partition (on /dev/sdb3) over to this drive. The layout is like so (after creating/formatting a 250GB partition on /dev/sdc):
What I would like to do is to copy /dev/sdb3 over to /dev/sdc1. So I booted up the Ubuntu install on /dev/sdb5, mounted sdb3 and sdc1, and then tried the following:Code:Disk /dev/sda: 400.0 GB, 400088457216 bytes
255 heads, 63 sectors/track, 48641 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x2b7a2b79
Device Boot Start End Blocks Id System
/dev/sda1 * 1 48640 390700768+ 7 HPFS/NTFS
Disk /dev/sdb: 250.0 GB, 250059350016 bytes
255 heads, 63 sectors/track, 30401 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x000ebf86
Device Boot Start End Blocks Id System
/dev/sdb1 * 1 8355 67111506 83 Linux
/dev/sdb2 8356 8486 1052257+ 82 Linux swap / Solaris
/dev/sdb3 8487 24421 127997887+ 83 Linux
/dev/sdb4 24422 30401 48034350 5 Extended
/dev/sdb5 24422 30401 48034318+ 83 Linux
Disk /dev/sdc: 500.1 GB, 500107862016 bytes
255 heads, 63 sectors/track, 60801 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x93c8bb99
Device Boot Start End Blocks Id System
/dev/sdc1 1 30394 244139773+ 83 Linux
It doesn't work, and the terminal output shows the following:Code:$ cd /media/sdb3
$ tar cf - . | ( cd /media/sdc1; tar xf -)
It displays this same message for every file on the entire sdb3 partition. I'm not sure why it would say that either -- clearly the files exist! Did I miss something here?Code:tar: ./var/lib/dpkg/alternatives/javaws: Cannot open: No such file or directory
tar: ./var/lib/dpkg/alternatives/jstat: Cannot open: No such file or directory
tar: ./var/lib/dpkg/alternatives/jmap: Cannot open: No such file or directory
tar: ./var/lib/dpkg/alternatives/ControlPanel: Cannot open: No such file or directory
I couldn't fault your commands and the method is what I would use normally.
However if your 250Gb disk sdb is failing then it is possible that filing access will be impeded by the corruption developed.
Since sdc is 500Gb and larger than the 250Gb of sdb a btter strategy is to clone the entire disk by
AFter the cloning you remove the sdb and hook the 500Gb disk at its position.Code:dd if=/dev/sdb of=/dev/sdc bs=32256
This arrangement has the following advantages
(1) The 500Gb will function exactly if not better than the 250Gb. The better comes from no corruption as the damaged areas are just bad data instead bad hardware.
(2) The new 500Gb will boot normally and any undamaged partitions will function perfectly. Thus you can use Ubuntu same as before.
(3) You test the distressed Fedora. If it boots then the damaged is not serious. In any case you can execute repair with it.
(4) The original sdb is a backup.
(5) You can use gparted to expand the partitions later on.
This scheme does require more work if you resize the partitions but you should be aware of the fact if sdb has corruption the whole disk will eventually go south. Also instead of resizing sdb3 you can always make new sdb6 or sdb7 of any size you want and migrate sdb3 to there without resizing any partition.
Well I was actually planning on just trashing Fedora. It was a Fedora 11 install that I had just put on there to play with, so there isn't much on there. I guess I could copy just the partition instead of the entire drive. So right now sdb3 is a 128GB partition and sdc1 is a 250GB partition. Will this work as-is without needed to expand the partition, or will I need to pop in gparted and expand sdc1 after copying?
If it needs expanding, does that mean that I need to destroy sdc1 and recreate it as a 128GB partition, clone sdb3, and then expand? Or is it acceptable to clone it over to the 250GB sdc1 that already exists and then just expand?Code:dd if=/dev/sdb3 of=/dev/sdc1 bs=32256
If you clone a partition or the entire disk using dd then you must remember it is a hardware type copying. In it the cloning procedure is controlled by counting sector by sector. This effectively means you can only use dd if the target disk is bigger than the source disk. And in the case of partition-to-partition cloning you need to ensure both partitions are indentical in size, down to the exact number of sectors. Disk-to-disk cloning is a lot saimpler and easier because there is nothing to do except making sure the target is exactly the same size or just larger.
Therefore dd should only be used for sector-by-sector cloning.
If you use tar then it is "the filing system" that get copied so the size of the partition is inmaterial! You do have to format the partition first before you can use tar. For dd you don't even need to bother with the partition table, let alone formatting each partition. So the two are totally different methods.
Linux can be copied from partition to partition without problem as long as you know how to mount the partitions using /etc/fstab and updating the Grub's menu.lst if the partition is used for booting.
I would not recommend you to use dd if the partition content can be migrated using tar. dd is good if you want simplicity (like cloning the whole disk) and the disk is actually having a read/write error (dd does not deal with the files, it copies the binary bits). If you clone the entire disk then the existing sdc1 will be overwritten.
It should be noted with respect to hard drive catastrophies:
Each time you power it up...may well be the last time you successfully do so
With that in mind you MUST have an effective plan to immediately rescue your most valuable data FIRST
This involves the following:
Do not allow the drive to be energized until you are absolutely ready
Do not allow others to accidentally energize the drive in your absence
Do not EVER boot from that hard drive again(until everything of value is rescued at least)
When you do energize the drive be prepared to immediately go directly to the most valuable data first and get it copied to a safe location immediately
Do NOT attempt to copy the whole drive
GO GET YOUR VALUABLE DATA FIRST!
A large portion of the data on the drive is of absolutely no serious value
You can easily replace the operating system and each and every piece of additional software and/or driver
Again, operate under the assumption that the drive will only last another ten minutes
Go to your most irreplaceable data FIRST and rescue it immediately
The biggest mistake we see at the service counter and technician's bench is making the assumption that the drive is going to stay operational through ANY specific process
As the owner of the equipment, consider that each time you turn your machine on...that may be the last time the system operates properly and without any losses
Hence our worldwide web battlecry...BACK IT UP! SECURE YOUR DATA!
And for the technician and hobbyist the same holds true and is even MORE important
You must demand that your friends and patrons give you a list of EXACTLY what must be rescued
You must demand BOTH what must be rescued AND the order of value(and hence the order in which data will be rescued)
You must proceed ONLY after an effective and complete plan of action is determined
Otherwise you will either be physically transplanting your platters into another hard drive to retrieve the data...
Or...
You will be paying someone else big money to do it for you
Again...
You have been warned!
So...
While your hard drive is operating perfectly fine, start backing-up anything and everything that you can't afford to lose FOREVER
As always, your mileage may vary, buyer beware, and buyer be aware
We now return you to your regularily scheduled programming
.
LewRockwell -- what a post!
I will say that I have already backed up all of my important stuff already, so I definitely have that covered. Thanks for the advice though. ;)
Quote:
Originally Posted by gamblor01
Just a heads up for Saikee and everyone else, this appears to have failed because I was trying to sudo it like so:
This failed with the errors I mentioned above. On a whim I thought that perhaps running sudo -i and then retrying the command would work...IT DID! So make sure you are truly root and not just trying to run the command via sudo.Code:$ sudo tar cf - . | ( cd /media/sdc1; tar xf -)
So, the complete set of steps I ran to copy all files from /dev/sdb3 to /dev/sdc1 was:
Code:$ sudo -i
# mkdir /media/sdb3
# mkdir /media/sdc1
# mount /dev/sdb3 /media/sdb3
# mount /dev/sdc1 /media/sdc1
# cd /media/sdb3
# tar cf - . | ( cd /media/sdc1; tar xf -)
I have been checking it and it is definitely copying everything over. I might need to reinstall grub on the drive (though the tar command should be copying over all of the /boot/grub directory, though my UUID will change as will the partitions), and I also need to create a swap partition. Then I should be able to disconnect sdb and put sdc in its place, boot it up, and hope for the best! I'll post an update later today.