help recovering HD

[bob smith_6]

I bought a pavilion 7935 almost exactly 1 year ago today (wonder what the warrantee is?) and installed Red Hat 7.3. It is a normal/typical install and sits on the internet, running apache. I've been keeping up on the security patches. Also, I recently installed samba so I could temporarily save off some data from a different machine while I reformatted it. Unfortunately, it seems that the HP's HD has crashed. It will boot up until it goes to mount some part of the FS and wants to do a fsck, but fails a read and the kernel panics. I used Tom's rtbt floppy and can see that the first 2 partitions are clean. But about 4 others are not even mountable.

I don't have the partition table listing (I'm not at home now) but it seems that it may be corrupt. I'm looking for advice on how I can recover the data off this drive and to see if the drive is bad.

Please let me know what info I can provide to help.

I going to try to dd the data to a new HD, but how do I do that if the HD's are different sizes? And, the source drive has server al partitions... do I mount those separately on the new HD once the old one is copied?

Thanks!
Bob

[Admin32]

I haven't tried this with Linux as yet, but it might work:

If you have another system with Linux on it, then simply connect the faulty hard drive to it and try to mount it after the system has booted. If your lucky, you might be able to get access to your data and then copy them straight to the system your using.

If you haven't got another system, then look for a small hdd to temporarely install linux on it and try the above.

I hope this helps !

Cheers,
Chris P.

[bob smith_6]

I am able to boot from a floppy, so I am able to run e2fsck and try to mount the partitions. Of the 7 partitions, only the first one (the boot one) passes the e2fsck and mount test. The rest fail on at least the e2fsck with a bad read of a sector and block. These are consistently the same blocks reported, though they are different ones for different partitions.

Separately, it seems that the partition table is messed up:

device boot start end blocks ID
/dev/hda1 * 1 7 528884+ 83
/dev/hda2 8 2912 21961800 83
/dev/hda3 2913 5182 17161200 83
/dev/hda4 5183 5295 854280 f
/dev/hda5 5183 5234 393088+ 83
/dev/hda6 5235 5269 264568+ 83
/dev/hda7 5270 5295 196528+ 82

Notice how partitions 4, 5, and 6 seem to overlap! Also note that partition 4 is set as a WIN95 Ext'd (LBA) partition. I never set it that way.

Also, it seems that copies of the superblock aren't working to restore that.

If I'd have to guess, I'd say that some process (perhaps Samba?) corrupted the FS. Any other help on getting data from this drive, please??

I have another HD that I could install and move data, but I don't know how I can if I can't mount the partitions.

Thanks,
Bob

[bob smith_6]

Block count on the first partition has an extra 4 at the end. The correct entry is:

/dev/hda1 * 1 7 52888+ 83

I'm thinking maybe I can guess at what the correct partition info should be from the block count?

Thanks,
Bob

[John Meissner]

Another option would be this - a bootable linux business card - it's a CD that will not install... but boot up using the CD and memory to linux - where you can work on the hard drive or copy files off the system.


http://www.linux.ie/download/ilug_bbc/
or
http://lbt.linuxcare.com/download.epl

are two such downloads sites

[bob smith_6]

Booting from CD seems about the same as booting from floppy. I can not properly access this drive. For example, when I did this:

e2fsck -nv /dev/hda2 > file 2>&1

I get this output:

e2fsck 1.27 (8-Mar-2002)
Pass 1: Checking inodes, blocks, and sizes
Error reading block 65541 (Attempt to read block from filesystem resulted in short read) while doing inode scan. Ignore error? no

Error while scanning inodes (32704): Can't read next inode


(This is only stdout. I can't seem to get stderr to show up in the file?)

Any other ideas how to get the data off? Or fix the drive?
Bob

[Paul Doyle_1]

The problem is probably with the HDD itself. Chances are it's a Fujitsu 40 GB drive. Much of HP's Pavilion product line made from 2000 to 2001 had various models of Fujitsu hard drive, most of which came from the factory with defective chips on their controllers. The defect causes the controller to forget what it's supposed to be doing, and the drive simply falls over, permanently. It happened to me, with my 7935. HP has a drive recall program. I think it's still in effect, and they will replace the drive free of charge.

[Huc_1]

When you dd a partition one an other disk , this only works if the partition is at least equal to or bigger then source partition.

I you would also try from /dev/hda6 up and do a backup of good partition you find.

In the end You may have to accept that you have lost a parttion !

I also noticed that you say no to correct error on fsck

I have reread man e2fsck and did you try e2fsck -p from hda6 upwards (BEWARE -p will try and correct) as oppose to -n that does inform (report bad blocks) you but does not do correction on your file system.

Keep us informed

Hope this helps

J-P