October 19, 201015 yr Hello all, I think I'm in need of some serious help . I was recently encountering lock-ups and then artifacting when watching videos in between (during the periods when I was able to keep the server up). After some time, I encountered some files which I could not delete. After some research on this forum and wiki, I ended up running reiserfchk --check on my drives (I have 4 data drives) since the issue appeared to be problems with random/individual files. md1 had some fixable errors. I ran --fix-fixable on that drive md2 was fine md3 was fine md4 suggests to run --rebuild-tree. Does anyone have suggestions on how I should proceed? From what I've read, it was suggested to get the help of this forum before proceeding since this was a "last-resort" option. Here's the output of the --check on md4: Replaying journal: Done. Reiserfs journal '/dev/md4' in blocks [18..8211]: 0 transactions replayed Checking internal tree.. \/ 1 (of 15|/ 73 (of 150-/110 (of 170|bad_indirect_item: block 334049437: The item (5 96 0x3af89001 IND (1), len 4048, location 48 entry count 0, fsck need 0, format new) has the bad pointer (866) to the block (334050390), which is in tree already / 5 (of 15// 65 (of 150//126 (of 170\bad_indirect_item: block 35946498: The item (5 324102 0x2be6d001 IND (1), len 4048, location 48 entry count 0, fsck need 0, format new) has the bad pointer (606) to the block (35945668), which is in tree already / 70 (of 150|/ 23 (of 170\bad_indirect_item: block 36850898: The item (5 324104 0x35fae001 IND (1), len 4048, location 48 entry count 0, fsck need 0, format new) has the bad pointer (290) to the block (36851507), which is in tree already bad_indirect_item: block 36850898: The item (5 324104 0x35fae001 IND (1), len 4048, location 48 entry count 0, fsck need 0, format new) has the bad pointer (430) to the block (36851647), which is in tree already /123 (of 150\bad_internal: vpf-10320: block 66672050, items 93 and 94: The wrong order of items: [77 324107 0x29db9001 IND (1)], [40 41 0x17001 IND (1)] the problem in the internal node occured (66672050), whole subtree is skipped finished Comparing bitmaps..vpf-10640: The on-disk and the correct bitmaps differs. Bad nodes were found, Semantic pass skipped 1 found corruptions can be fixed only when running with --rebuild-tree
October 19, 201015 yr Basically you are on the right path. Follow its instructions and invoke it with the rebuild-tree option. It will create a lost+found folder and put in it the files and parts of files it cannot identify. Joe L.
October 19, 201015 yr Author Awesome!!! Thank you very much for the quick reply Joe! I very much appreciate it. At this point, is it worth it to attempt a backup of the drive prior to the rebuild? or would that be pointless? thx! -alex
October 20, 201015 yr Awesome!!! Thank you very much for the quick reply Joe! I very much appreciate it. At this point, is it worth it to attempt a backup of the drive prior to the rebuild? or would that be pointless? thx! -alex You could, if you like... but it might have to be an image copy vs specific files, since the file-system is corrupted.
October 20, 201015 yr Author So I let the --rebuild-tree run overnight without backing up (figuring the worst that could happen would be that I would need to rip some DVDs again). It completed, but what's strange (at least to me) is that there weren't any files in the lost+found directory. It did say that it corrected a bunch of files, and deleted 2 "unreachable" ones, so maybe I should count myself lucky with only 2 files I need to contend with? Anyways, I'm now wondering what the recommended next steps would be? Once the tree is rebuilt, is the filesystem "good" again (and by good, I mean no need to do anything else at this point, besides, for example, delete my empty lost+found directory)? Or should I consider it "fragile" and therefore preclear and rebuild using parity or backup? Speaking of parity, should I trust it, since so many files were touched on md1 and md4? And here's the final results of the tree rebuild, in case anyone's curious. Flushing..finished Files found: 315935 Directories found: 8213 Files with fixed size: 1 Pass 3a (looking for lost dir/files): ####### Pass 3a (lost+found pass) ######### Looking for lost directories: Flushing..finished90, 36 /sec Pass 4 - finished done 332538, 26 /sec Deleted unreachable items 2 Flushing..finished Syncing..finished ########### reiserfsck finished at Wed Oct 20 05:03:06 2010
October 20, 201015 yr It is now good. No need to clear/format it again. If you wish to be certain, run the "--check" once more, but I'm pretty sure it will find nothing. As far as the lost+found, although the contents may not have their original names you might be able to figure out what they used to be named and re-name them. Their size might be a clue. Joe L.
October 20, 201015 yr Author Thanks again Joe, Oddly enough, my lost+found didn't have any contents ("ls -l" gives me total 0). I guess that means I was either lucky, or need to figure out which files are missing (it said that it just deleted 2, i think) Regardless, I'm just happy that I'm almost done with this recovery effort and getting my server back into tip-top shape . At the suggestion of some other posts in this forum, I will do a memtest after work tonight just to eliminate that as a possible culprit. I did one when building my server, but didn't run it overnight/for a few good hours. thx so much! -alex
Archived
This topic is now archived and is closed to further replies.