Re: [Dar-support] Re: Input/output error
For full, incremental, compressed and encrypted backups or archives
Brought to you by:
edrusb
From: Johnathan R. <jr...@be...> - 2006-03-30 20:43:29
|
Thanks again for your help. I have a ReiserFS filesystem, not ext2/3, so I need to use reiserfsck (the way I understand it). I don't see any obvious option in the documentation to have that check for bad blocks, and the ReiserFS site seems to imply you need to use the badblocks program manually. Should I try "badblocks -n", then? Is this definitely going to be non-destructive, I don't see how you can read and write to/from every block and not lose data... And I'm still a little unclear on how you're saying I should test the entire backup after trying diff on each slice. Are you saying copy all the slices from CD back to the bad hard drive and test again? Denis Corbin wrote: > Johnathan Ritzi wrote: >> I did a "dar -t" on the copies on the hard disk, and I get the following: >> >> "ERR <ROOT>/home/myuser/.thunderbird/default.bnf/Mail/News & >> Blogs/CalStuff : compressed data CRC error >> > > this means the archive is corrupted and that corruption concerns this > file (it will not be possible to recover it). All other files should be > OK. Note that the archive itslef seems to be stored on blocks among > which one is "bad". If the test has been performed on the definitive > media (CD/DVD for example), it should be OK to make a new backup only > for the file with error and *test* the archive once wrote on a safe place. > > [...] > >> >> Does this indicate a problem with just this file (it is in the same >> directory as the bad one), or corruption that will affect the entire >> backup? The "cat" test on that file shows nothing wrong with it. > > the file itslef is problably sane, this is the archive part that holds > its data that is corrupted. > >> >> At this point, I'm burning my final backups to CD, one slice per CD. >> Is there a way to check the integrity of the backup one slice at a >> time from the mounted CD, or is the only way to really be sure to copy >> them all back to the hard drive and test them all at once? > > you can use diff to compare the burnt slice with the one you have on > disk. Then it is not bad to make a global test with dar -t on your fresh > set of CD (this is the only way to be sure that dar will be able to read > your archive, for example, that no slice is missing). > >> >> As far as the bad sectors go, I'm all for replacing my drive (which is >> covered under warranty), but the manufacturer will only take it back >> if I can show with some diagnostic tool that I do indeed have bad >> blocks. The problem is, I run the "badblocks" utility, and it exits >> without any output, indicating I have no bad blocks. I'm a little >> confused since everyone seems to say the dma error definitely means >> there are bad sectors, but the utility designed to detect them is >> telling me I'm fine... > > the badblock utility has two modes of functionning. Quoted from > badblocks man page: > -n Use non-destructive read-write mode. By default only a > on-destructive read-only test is done. > [...] > -w Use write-mode test. With this option, badblocks scans for > bad blocks by writing some patterns (0xaa, 0x55, 0xff, 0x00) on every > block of the device, reading every block and comparing the contents. > This option may not be combined with the -n option, as they are mutually > exclusive. > > Note that "badblocks -w" is a *destructive test* it will erase your > partition. Prefer the "e2fsck -c -c option". Quoted from e2fsck man page : > > -c This option causes e2fsck to run the badblocks(8) program > to find any blocks which are bad on the filesystem, and then marks them > as bad by adding them to the bad block inode. If this option is > specified twice, then the bad block scan will be done using a > non-destructive read-write test. > > > Regards, > Denis. |