Hi there, I started getting errors on one of my 2 drives, so I got a warranty one. Then I made an exact copy of the failing one onto the new one with dd, and swapped them. I ran fsck on all the affected partitions, and things looked good. Then I started using the striped RAID device (ext3) which had some minor data loss (mostly under big files), and I got the problems with the journal. The error I got was "journal block not found at offset 7180". The device switched itself into read-only mode. There were no read errors, just the problems due to the fact that the copy of the corrupted drive was slightly corrupted as well. I ran e2fsck, but of course it did not fix anything. I encountered the problem a second time. After googling, I figured I needed to recreate the journal. So I used (tune2fs, e2fsck, tune2fs) sequence to remove the journal and then add it. Things appear to work fine after that. Since tune2fs takes negligible time as opposed to e2fsck, should not ext3->ext2->ext3 sequence be included into the e2fsck call? That would take care of the case of corrupted journal but good filesystem. At least, when the failure was caused by the journal and not the system itself ... Konstantin __________________________________ Do you Yahoo!? Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html
Konstantin> I started getting errors on one of my 2 drives, so I got Konstantin> a warranty one. Then I made an exact copy of the failing Konstantin> one onto the new one with dd, and swapped them. How were these two disks setup, in a RAID0 stripe? So you have one filesystem spread across both disks, correct? With the cost of disks today, I'd strongly recommend that the only RAID you use is RAID1 (mirroring) or RAID5. Don't strip across disks, because when one disk fails, or starts to fail, you're in for a world of pain. I consider you lucky to have gotten back as much data as you did, but I'd also worry that your remaining data is suspect, esp since fsck just checks the meta-data of the filesystem, but does nothing to check whether the data inside various files is actually any good. So you may have even more corruption hitting you down the line here, as you access and check files. Just my two cents, from a man who runs a pair of mirrored 120gb HD for his /home and /local volumes.
Konstantin Kudin <konstantin_kudin@yahoo.com> wrote:> [ errors on hard disk, dd, fsck, problems with the > journal, e2fsck, it did not fix anything, recreate > the journal] > > should not ext3->ext2->ext3 sequence be > included into the e2fsck call?1. IMHO yes 2. This should be a Feature Request @ SF.net: http://sourceforge.net/tracker/?func=add&group_id=2406&atid=352406 Regards, Bodo -- Mike: Could my system get instable, if I install this program? Cole: Your system will get instable in any case, so it doesn't matter.