CompuServe Thread

#Hard Disk Crash

12 messages in this thread
#100045From: Greg WhiteDec 26, 1987 12:09 AM
Don, Well, after 9 months I finally had my first HD crash. I had a power glitch (luckily, very rare) while writing to my CLtd HD. The glitch was fast enough so that my Amiga did not lose its mind, but my disk drive sat there and clucked to itself for a while with the LED flashing for about 5 minutes and then settled down. I was then able to read and write to the HD with no problems. To be sure I had not corrupted the system, I rebooted. I was promptly informed my HD was no longer a DOS disk. As an experiment I decided to try Diskdoctor (I know, I know, but I was curious). We had a similar problem with our PAL Sr. drive at work and were told by BYTE by BYTE to use Diskdoctor to ressurect it. If you haven't tried it, it renames your HD Lazarus. Nice touch. With our PAL it worked great. When I tried it on my CLtd, it didn't work quite so well. It read all 600 and something tracks and then started replacing files. It came with 7 files it listed as corrupted and ask if I wanted to delete them. I did. All seem to be OK reading the drive, however when I tried to write to it, Amy said the drive was not validated. When I did an info from CLI, it said the drive was validating. I then rebooted and the drive spent about 5 mins during the s-s and then gave me a system requestor saying "Error validating disk Key 13532 – bad extension" When I clicked "cancel" another one came up and told the "disk structure corrupt. Use diskdoctor" After this, I could read the HD but still not write to it. Info says it is validating. However after waiting overnight, it was still validating. I repeated using diskdoctor and got exactly the same results. Of course I can't use Dave Haynie's Disksalvage without another 24 meg HD. Out of curiousity I still would like to see if I can recover the drive without reformatting. Any ideas on how to get the drive to validate? Anyone else have any ideas? Later, Greg
#100050From: Don Curtis/SYSOPDec 26, 1987 12:59 AM
Greg, A few ideas…both of which have worked for me during testing: 1. Look for a 0 length file (more than likely, the file you were writting at the time of the crash) and delete it. That may or may not help. 2. Copy the files one at a time until you get a system error on a file and delete that file. Now…I know that #2 sounds like an awful lot of work…depending on how much RAM you have and how your HD is organized…it can be very easy to do. If you have a relatively well organized HD and no huge files on it and at least 1 Meg of RAM, then open a 2nd CLI via the NEWCLI command. In that new CLI, cd to ram: and do likewise in your original CLI. Ok, now from one of the CLI's (it doesn't matter which) type the following: COPY DH0: RAM: ALL This will go thru and attempt to copy the HD into RAM: and create subdirectories as it goes. MAKE NOTE of which directories and subdirectories are being created…it's important! Obviously, soon, you'll get a system requester telling your that ram: is full. Not to worry….go to the other CLI and delete the directores that have already been copied using the DELETE <directory> ALL command. The reason you need to keep track of which directories and/or sub-directories are being copied is that you won't be able to delete the directory that it is presently trying to write to when ram: filled up and the DIR command won't work once RAM: gets full! There is no RAM left for DIR to use to sort. Ok, once you've deleted some directories and/or subdirectories (get as many as you can) then hit the RETRY gadget in the system requester telling your RAM: is full (it's not now). Also click on the CLI window that is doing the copies. As long as you have no individual files that are bigger than RAM can hold…this will work. At some point, you'll come across a file that can't be copied due to the bad extension….that's the file you want to delete. Once you delete that file…and after waiting a reasonable amount of time for the drive to quit doing it's thing….re-boot and the drive should now validate. Don
#100151From: Greg WhiteDec 26, 1987 11:15 PM
Don, I appreciate all your help in this. I have found the problem. Apparently there is a bad file in one particular directory. When ever I try to list this directory in any way whatsoever, I get a "disk error – task held" GURU. No matter what, I just get the first 39 files and then the system locks up. Unfortunately I have no idea what the name of the bad file is, and I see no way to find out. So I can't try to delete it. I guess I could try to "delete#?", but was hoping to avoid this. Any more ideas? Later, Greg
#100153From: Don Curtis/SYSOPDec 26, 1987 11:39 PM
Greg, Yup…try what I said before, try copying that directory to ram: and as it's doing the copy, it will list the filenames one at a time, you'll see the filename and then get the guru. copy <directory> ram: all That should do it. Once you find the filename, then delete <directory>/filename and hopefully that should take care of it. Again, you'll probably have to reboot to get the drive to reverify. But this time it should go ok. Remember to do the copy from a separate CLI so that you can go to RAM: and delete files if RAM: fills up and then hit the RETRY gadget in the system requester that will be displayed if RAM: fills up. Try it and let me know how it works, it _should_ work ok. Don
#100158From: Steve AhlstromDec 27, 1987 12:04 AM
Don, Nope. Won't work. It's reformat time.
#100157From: Steve AhlstromDec 27, 1987 12:03 AM
Greg, Just ran across the same problem last nite. Had a file go bad. When I try to access the file in anyway, GURU time. Solution? Grab your Saf-T-Net, do a backup (full or incremental) excluding that file. Format your HD, then restore from the back. That's the only solution. IF the file went bad because of a hard error, you may have to reprep and map that sector out.
#100167From: John DraperDec 27, 1987 2:38 AM
Steve, In a case like that, i would use Sectorama, go into that directory, and unlink the bad file. It requires a bit of knowledge about the filing system, but it's all there in the Amigados manual from bantam (or the Amigados Tech Ref manual from CBM). Regards, Larry.
#100169From: Steve AhlstromDec 27, 1987 2:44 AM
Didn't think of that! Oh well, maybe next time. It reinforces the need to have a current backup, tho!
#100170From: John DraperDec 27, 1987 2:46 AM
Sure does Steve! There really is no subsitute for a current backup. A file unlinked is a file gone, unless you have it backed up. Regards, Larry.
#100259From: Lloyd W. Dull IIIDec 27, 1987 8:36 PM
Steve, How do you like Saf-T-Net, by the way? How does it compare to, lets say BRU2 or Steve Drew's Backup? Lloyd
#100268From: Steve AhlstromDec 27, 1987 9:35 PM
Saf-T-Net is a real HD backup program not a half hearted attempt like BRU2 and Backup. It's not as nice as some available in the MS-DOS world, but, I think it's the best currently available for Amiga.
#100555From: Lloyd W. Dull IIIDec 29, 1987 10:13 PM
Steve, Thanks. That's what I thought. Now if only someone could come out with a "Fastback" type of HD backup! Lloyd