Re: [Bug-ddrescue] How to properly repair rescued image?
On Mon, 11 Jun 2007, James W. Watts wrote: I used ddrescue v1.3 to copy a crashed Windows XP NTFS partition. The rescued NTFS image file is stored on a hard drive with a single EXT2 partition. Referencing the tutorial in 'ddrescue.info', I will next copy the image to a different drive for the repair stage. The tutorial states, After the copy is repaired, with e2fsck or some other tool appropriate for the type of partition... I'm confused. Does this mean that the tool for repair needs to match the file system of the rescued image file (NTFS in my case) or the file system of the partition on which the rescued image file resides (EXT2 in my case)? Please advise. If the latter is true, I presume I will be able to run e2fsck against the image for repair. If the former is true, I'm not sure how to repair the image. Would I need to copy the image to an NTFS-formatted hard drive, mount that HD in an XP machine, then run an NTFS partition repair tool? I've looked for such a tool that is Linux-based (e.g. ntfsck), but it does not yet exist. You need to create a partition EXACTLY the same size as the old partition was (down to the byte exactly). Then copy the ntfs data to that partition, boot windows and have windows checkdisk (or some other commercial tool) work on it. Unlike linux windows can not open a filesystem that is stored as a regular file, it has to be in a partition (and make sure the partition type is correct). Hmm, actually it's theoretically possible you could use VMware, or other virtualizer, to mount the file as if it was a real partition. When you create the partition you must use a tool that can show you exactly how big it is (in sectors - multiply by 512 to get bytes). You may need to tell the partitioning tool to use a specific number of heads and sectors-per-track (i.e. to use the same number as was used on the original partition), otherwise you will find it impossible to make a partition exactly the same size. Good luck. -Ariel ___ Bug-ddrescue mailing list Bug-ddrescue@gnu.org http://lists.gnu.org/mailman/listinfo/bug-ddrescue
Re: [Bug-ddrescue] Is there a ddrescue manual?
On Wed, 14 Mar 2007, ddMAC wrote: The fact that I was able to get a control line program up and running should Do you mean command line? (Not control line.) dd bs=512 if=/Volumes/Beatles of=/Volumes/Macintosh\ HD1/foofoo.dmg conv=noerror,sync To try the stop feature of ddrescue I stopped the terminal after You say ddrescue, but you are running dd, which one are you using? -Ariel ___ Bug-ddrescue mailing list Bug-ddrescue@gnu.org http://lists.gnu.org/mailman/listinfo/bug-ddrescue
Re: [Bug-ddrescue] ddrescue 1.3 - questions from a newbie
On Wed, 10 Jan 2007, Matt Boge wrote: PS. I think I'd like to consider creating multiple partitions in my new setup (two 400GB drives that I will be setting up with RAID0). Watch out with RAID0, if you do that, if either one of the drives die you lost the data from both. And recovery is much more complicated because half the data is on each drive. I'd install Windows XP on one partition and I think I'd like to install a Linux variant on the other to play around with and get more familiar with this OS (hey, an old dog can ALWAYS learn new tricks, right?). Any suggestions on how I should do this and which variant of Linux I should install? Create the partitions outside linux, first create the Windows partition (type doesn't matter, as you will see). Then create the Linux partition, and then finally delete the windows one. You are doing it in this strange order to make sure the second partition was also the second one created. Don't forget to create a swap partition if you will use one (you can also swap onto a data partition). Don't install linux yet (since potentially windows will erase it, and then you'll have wasted your time) boot windows setup, and ask to partition the drive - windows should complain about some mystery partition on the drive, but ignore that, and let window partition the free space on the drive. Finish windows install. Now install linux, first of all linux will not get confused about the extra partition, second the installer will (should) notice windows and create an entry for it. And finally linux will do the right thing in regard to making sure you can actual boot (window can't handle it). Anyway, as far as what variant (called a distribution) I like Debian, but try these pages: http://www.tuxs.org/chooser/ http://www.zegeniestudios.net/ldc/ -Ariel PS. Another option (if your chosen distib supports it), is to be dumb, and let windows own the whole drive, then have the distrib shrink the partition to make room for linux. If you use RAID on the drives you are complicating things, and I just don't know what linux will do. If it's hardware raid it should work, but if you use windows raid, I'm just not sure. ___ Bug-ddrescue mailing list Bug-ddrescue@gnu.org http://lists.gnu.org/mailman/listinfo/bug-ddrescue
Re: [Bug-ddrescue] Interpreting logfile of restore to get list of corrupted files
On Wed, 27 Sep 2006, Chris Witham wrote: On 9/26/06, Ariel [EMAIL PROTECTED] wrote: http://smartmontools.sourceforge.net/BadBlockHowTo.txt Thank you. This is exactly the information I wanted. The problem is that my failed partition is ReiserFS. The icheck and ncheck commands are part of debugfs, which is ext2/ext3-specific, and all I can find online is other people complaining that ReiserFS doesn't have an equivalent command. There doesn't appear to be any way to get ReiserFS to tell what files are on what sectors. You might want to email a list dedicated to ReiserFS. However, ddrsummarize.pl (thank you Dave!) shows that the 1688 bad sectors mentioned in the logfile actually translates to less than 1MB of data lost. 1688 * 512 = 800KB. This makes me happy (I can deal with 800kB of lost data, out of 111GB) and annoyed (all this trouble over only 800kB?). That's actually normal. And it's the best way to fail a hard disk - just a few sectors go bad, and you find out right away, before more fail. Be glad it's this type of failure, and not a total loss of the hard disk. BTW what model hard disk is this? Failures are inevitable, when a disk fails slowly like this it's a plus. I will just restore the drive and find out through trial and error what does and doesn't work. That won't work. Any file that was not restored will have 0's, for the data. It's not something that is easy to detect, it will not give a disk error, it just will have a blank area. If it's a program it might not even crash till it happens to execute that code, depending on your usage, it might never. A document will simply have a blank area, etc. If the 0's are for a directory entry, that should be pretty obvious when you run fsck, but for files you won't be able to tell without checking each one. You should copy your data, and reinstall the OS. If that's too much work, then at least refresh each file - meaning get the original, and copy it over the existing. (In debian it's easy, just apt-get --reinstall and a list of packages.) One other options: if you are 100% done with recovering your data, and you don't need anything else on the old hard disk, then try to mount it, and then copy every single file to /dev/null - then watch where the errors are. But be careful - the kernel could crash if the fs is corrupted badly. Also if a directory entry is bad, you will not be able to check the files under it. -Ariel ___ Bug-ddrescue mailing list Bug-ddrescue@gnu.org http://lists.gnu.org/mailman/listinfo/bug-ddrescue
Re: [Bug-ddrescue] Disc or partition?
On Thu, 3 Aug 2006, David P James wrote: On Thu 3 August 2006 13:06, Ariel wrote: You wrote that fdisk can't read the partition table, yet you mention hdb1 and hdb5? I know the partition structure from memory (one primary and one logical in extended), not from fdisk, which cannot read what I know to [have] be[en] there. I hope that clears that up. Yes, you don't actually have hdb1, and hdb5. So go with my advice here: However, if linux can't read the partition table of hdb, then you have no choice, but to copy the entire disk. Then later run a tool that will attempt to rescue your partition, by searching the disk for 'start of filesystem'. Linux cannot read the partition table at all of hdb (only 'hdb' shows up in cat /proc/partitions and that's giving funny numbers, the equivalent of something like 137 GB total). disk, or should the new disk be left blank and the creation of partitions to be handled (hopefully) by a rescue tool after the fact? Yes. This one: ddrescue [options] /dev/hdb /dev/hdc rescue.log ? VERY IMPORTANT: zero out hdc first!! If you leave any data on hdc, then any tools you run will 'see' that old data and get very confused. Many people recommend using a disk-image, rather then a partition, or disk, for this reason. And good luck with the recovery. I suspect the data on the disk may be completely unrecoverable but I'd like to rule it out completely before doing a partial restore from an 8 month old image (of C: drive only). Sorry to hear that, data losses really suck, it happened to me, but I was able to recover about 90%. -Ariel ___ Bug-ddrescue mailing list Bug-ddrescue@gnu.org http://lists.gnu.org/mailman/listinfo/bug-ddrescue
Re: [Bug-ddrescue] Recover a single NTFS partition off a damaged disk using ddrescue
On Fri, 28 Jul 2006, Sanjay Rana wrote: Is it possible to recover a single NTFS partition as a disk image onto a different hard disk using ddrescue? If so then what arguments will have to be used to do that and will it be possible to retrieve the data from the disk image? My hard disk has lots of bad sectors in other partitions and I am unable to overwrite the partition table. Mirroring the entire disk is also taking a lot longer as well. I know the start and end of the partition in terms of the cylinders,head,sector as reported by testdisk. Can this information be useful? What you will need to do is create a partition on the good disk that exactly matches the size (in sectors!) of the old one. It is very important that the sector size matches. (Easiest way to do this, is partition the new hard disk using the same heads and sectors per track as the old one. You can only do this in linux, windows doesn't let you choose these numbers (although you might get lucky and they'll be the same anyway). Then create a partition with the same number of cylinders as the old one.) Then use ddrescue to recover to that partition. You could use a disk image, rather then a partition, but windows will not be able to use it, only linux. I think windows has better tools for recovering corrupted NTFS file systems, so I suggest a partition. Actually - you could maybe use vmware, with a disk image. It will trick windows into thinking it's a normal partition on a hard disk. It's a little complex, but probably doable. BTW: If you get anything useful from that partition, I suggest copying all the data off of it, to a new area. Don't keep the partition afterward, since it was the result of a corrupted NTFS, rather then a good format. Basically you will need double the amount of space of the bad partition. Hope this helps. -Ariel ___ Bug-ddrescue mailing list Bug-ddrescue@gnu.org http://lists.gnu.org/mailman/listinfo/bug-ddrescue